A new release 0.4.14 of RQuantLib was uploaded to CRAN earlier today, and has by now been uploaded to Debian as well. QuantLib is a very comprehensice free/open-source library for quantitative finance; RQuantLib connects it to the R environment and language. The release of RQuantLib comes just one months after the previous release, and brings three changes. First, we added both two more US-based calendars (including FederalReserve ) along with a bunch of not-yet-included other calendars which should complete the coverage in the R package relative to the upstream library. Should we have forgotten any, feel free to open an issue. Second, CRAN currently aims to have older autoconf conventions updated and notified maintainers of affected packages. I received a handful of these, and just like yesterday s update to littler refreshed this here. Third, we set up automated container builds on GitHub. No other changes were made, details follow.

Changes in RQuantLib version 0.4.14 (2021-10-06)

• Changes in RQuantLib code:
  • Several new calendars were added (Dirk in #159 closing #155)
• Changes in RQuantLib package and setup:
  • Docker containers are now updated on a monthly schedule via GitHub Actions
  • The configure files were updated to the standard of version 2.69 following a CRAN request

Courtesy of my CRANberries, there is also a diffstat report for the this release. As always, more detailed information is on the RQuantLib page. Questions, comments etc should go to the new rquantlib-devel mailing list. Issue tickets can be filed at the GitHub repo. If you like this or other open-source work I do, you can now sponsor me at GitHub.

This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. Please report excessive re-aggregation in third-party for-profit settings.

The fifteenth release of littler as a CRAN package just landed, following in the now fifteen year history (!!) as a package started by Jeff in 2006, and joined by me a few weeks later. littler is the first command-line interface for R as it predates Rscript. It allows for piping as well for shebang scripting via #!, uses command-line arguments more consistently and still starts faster. It also always loaded the methods package which Rscript only started to do in recent years. littler lives on Linux and Unix, has its difficulties on macOS due to yet-another-braindeadedness there (who ever thought case-insensitive filesystems as a default were a good idea?) and simply does not exist on Windows (yet the build system could be extended see RInside for an existence proof, and volunteers are welcome!). See the FAQ vignette on how to add it to your PATH. A few examples are highlighted at the Github repo, as well as in the examples vignette. This release updates the helper scripts to download nighlies of RStudio Server and Desktop to their new naming scheme, adds a downloader for Quarto, extends the roxy.r wrapper with a new option, and updates the configure setting as requestion by CRAN and more. See the NEWS file entry below for more.

Changes in littler version 0.3.14 (2021-10-05)

• Changes in examples
  • Updated RStudio download helper to changed file names
  • Added a new option to roxy.r wrapper
  • Added a downloader for Quarto command-line tool
• Changes in package
  • The configure files were updated to the standard of autoconf version 2.69 following a CRAN request

My CRANberries provides a comparison to the previous release. Full details for the littler release are provided as usual at the ChangeLog page, and now also on the new package docs website. The code is available via the GitHub repo, from tarballs and now of course also from its CRAN page and via install.packages("littler"). Binary packages are available directly in Debian as well as soon via Ubuntu binaries at CRAN thanks to the tireless Michael Rutter. Comments and suggestions are welcome at the GitHub repo. If you like this or other open-source work I do, you can now sponsor me at GitHub.

This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. Please report excessive re-aggregation in third-party for-profit settings.

Like each month, have a look at the work funded by Freexian s Debian LTS offering. Debian project funding In August, we put aside 2460 EUR to fund Debian projects. We received a new project proposal that got approved and there s an associated bid request if you feel like proposing yourself to implement this project. We re looking forward to receive more projects from various Debian teams! Learn more about the rationale behind this initiative in this article. Debian LTS contributors In August, 14 contributors have been paid to work on Debian LTS, their reports are available:

• Abhijith PA did 4.0h (out of 14h assigned and 5h from August), thus carrying over 15h to September.
• Adrian Bunk did 11h (out of 23.75h assigned), thus carrying over 12.75h to September.
• Anton Gladky did 12h (out of 12h assigned).
• Ben Hutchings did 1.25h (out of 13.25h assigned and 6h from August), thus carrying over 18h to September.
• Chris Lamb did 18h (out of 18h assigned).
• Emilio Pozuelo Monfort did not report back about their work so we assume they did nothing (out of 23.75h assigned plus 50.75h from August), thus is carrying over 74.5h for September.
• Holger Levsen did 3h (out of 12h assigned) to help coordinate the team, and gave back the remaining hours.
• Lee Garrett did nothing (out of 23.75h assigned), thus is carrying over 23.75h for September.
• Markus Koschany did 35h (out of 23.75h assigned and 30h from August), thus carrying over 18.75h to September.
• Neil Williams did 24h (out of 23.75h assigned), thus anticipating 0.25h of October.
• Roberto C. S nchez did 22.25h (out of 23.75h assigned), thus carrying over 1.5h to September.
• Sylvain Beucler did 21.5h (out of 23.75h assigned), thus carrying over 2.25h to September.
• Thorsten Alteholz did 23.75h (out of 23.75h assigned).
• Utkarsh Gupta did 23.75h (out of 23.75h assigned).

Evolution of the situation In August we released 30 DLAs.

This is the first month of Jeremiah coordinating LTS contributors. We would like to thank Holger Levsen for his work on this role up to now.

Also, we would like to remark once again that we are constantly looking for new contributors. Please contact Jeremiah if you are interested! The security tracker currently lists 73 packages with a known CVE and the dla-needed.txt file has 29 packages needing an update. Thanks to our sponsors Sponsors that joined recently are in bold.

• Platinum sponsors:
  • TOSHIBA (for 72 months)
  • GitHub (for 63 months)
  • Civil Infrastructure Platform (CIP) (for 40 months)
• Gold sponsors:
  • Roche Diagnostics International AG (for 83 months)
  • Linode (for 77 months)
  • Babiel GmbH (for 67 months)
  • Plat Home (for 66 months)
  • University of Oxford (for 22 months)
  • Deveryware (for 9 months)
  • VyOS
• Silver sponsors:
  • The Positive Internet Company (for 89 months)
  • Domeneshop AS (for 88 months)
  • Nantes M tropole (for 82 months)
  • Universit Jean Monnet de St Etienne (for 74 months)
  • Univention GmbH (for 73 months)
  • Ribbon Communications, Inc. (for 67 months)
  • Exonet B.V. (for 57 months)
  • Leibniz Rechenzentrum (for 51 months)
  • CINECA (for 41 months)
  • Minist re de l Europe et des Affaires trang res (for 35 months)
  • Cloudways Ltd (for 24 months)
  • Dinahosting SL (for 22 months)
  • Bauer Xcel Media Deutschland KG (for 16 months)
  • Platform.sh (for 16 months)
  • Moxa Intelligence Co., Ltd. (for 10 months)
  • sipgate GmbH (for 7 months)
  • Tilburg University (for 6 months)
  • OVH US LLC (for 5 months)
• Bronze sponsors:
  • Evolix (for 88 months)
  • Seznam.cz, a.s. (for 88 months)
  • Intevation GmbH (for 85 months)
  • Linuxhotel GmbH (for 85 months)
  • Daevel SARL (for 84 months)
  • Bitfolk LTD (for 83 months)
  • Greenbone Networks GmbH (for 82 months)
  • Megaspace Internet Services GmbH (for 82 months)
  • NUMLOG (for 82 months)
  • WinGo AG (for 81 months)
  • Ecole Centrale de Nantes LHEEA (for 77 months)
  • Entr ouvert (for 72 months)
  • Adfinis AG (for 70 months)
  • GNI MEDIA (for 64 months)
  • Laboratoire LEGI UMR 5519 / CNRS (for 64 months)
  • Tesorion (for 64 months)
  • Bearstech (for 56 months)
  • LiHAS (for 56 months)
  • People Doc (for 52 months)
  • Catalyst IT Ltd (for 50 months)
  • Demarcq SAS (for 44 months)
  • Universit Grenoble Alpes (for 30 months)
  • TouchWeb SAS (for 22 months)
  • SPiN AG (for 19 months)
  • CoreFiling (for 15 months)
  • Institut des sciences cognitives Marc Jeannerod (for 9 months)
  • Observatoire des Sciences de l Univers de Grenoble (for 6 months)
  • Tem Innovations GmbH
  • WordFinder.pro

Armadillo is a powerful and expressive C++ template library for linear algebra aiming towards a good balance between speed and ease of use with a syntax deliberately close to a Matlab. RcppArmadillo integrates this library with the R environment and language and is widely used by (currently) 912 other packages on CRAN. This new release brings us Armadillo 10.7.0 released this morning by Conrad. Leading up to this were three runs of reverse dependencies the first of which uncovered the need for a small PR for subview_cols support which Conrad kindly supplied. The full set of changes follows. We include the last interim release (sent as usual to the drat repo) as well.

Changes in RcppArmadillo version 0.10.7.0.0 (2021-09-30)

• Upgraded to Armadillo release 10.7.0 (Entropy Maximizer)
  • faster handling of submatrix views accessed by X.cols(first_col,last_col)
  • faster handling of element-wise min() and max() in compound expressions
  • expanded solve() with solve_opts::force_approx option to force use of the approximate solver

Changes in RcppArmadillo version 0.10.6.2.0 (2021-08-05)

• Upgraded to Armadillo release 10.6.2 (Keep Calm)
  • fix incorrect use of constexpr for handling fixed-size matrices and vectors
  • improved documentation
• GitHub- and drat-only release

Courtesy of my CRANberries, there is a diffstat report relative to previous release. More detailed information is on the RcppArmadillo page. Questions, comments etc should go to the rcpp-devel mailing list off the R-Forge page. If you like this or other open-source work I do, you can sponsor me at GitHub.

This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. Please report excessive re-aggregation in third-party for-profit settings.

SReview, the video review and transcode tool that I originally wrote for FOSDEM 2017 but which has since been used for debconfs and minidebconfs as well, has long had a sizeable component for inspecting media files with ffprobe, and generating ffmpeg command lines to convert media files from one format to another. This component, SReview::Video (plus a number of supporting modules), is really not tied very much to the SReview webinterface or the transcoding backend. That is, the webinterface and the transcoding backend obviously use the ffmpeg handling library, but they don't provide any services that SReview::Video could not live without. It did use the configuration API that I wrote for SReview, but disentangling that turned out to be very easy. As I think SReview::Video is actually an easy to use, flexible API, I decided to refactor it into Media::Convert, and have just uploaded the latter to CPAN itself. The intent is to refactor the SReview webinterface and transcoding backend so that they will also use Media::Convert instead of SReview::Video in the near future -- otherwise I would end up maintaining everything twice, and then what's the point. This hasn't happened yet, but it will soon (this shouldn't be too difficult after all). Unfortunately Media::Convert doesn't currently install cleanly from CPAN, since I made it depend on Alien::ffmpeg which currently doesn't work (I'm in communication with the Alien::ffmpeg maintainer in order to get that resolved), so if you want to try it out you'll have to do a few steps manually. I'll upload it to Debian soon, too.

Review: The Problem with Work, by Kathi Weeks

Publisher:	Duke University Press
Copyright:	2011
ISBN:	0-8223-5112-9
Format:	Kindle
Pages:	304

One of the assumptions baked deeply into US society (and many others) is that people are largely defined by the work they do, and that work is the primary focus of life. Even in Marxist analysis, which is otherwise critical of how work is economically organized, work itself reigns supreme. This has been part of the feminist critique of both capitalism and Marxism, namely that both devalue domestic labor that has traditionally been unpaid, but even that criticism is normally framed as expanding the definition of work to include more of human activity. A few exceptions aside, we shy away from fundamentally rethinking the centrality of work to human experience. The Problem with Work begins as a critical analysis of that centrality of work and a history of some less-well-known movements against it. But, more valuably for me, it becomes a discussion of the types and merits of utopian thinking, including why convincing other people is not the only purpose for making a political demand. The largest problem with this book will be obvious early on: the writing style ranges from unnecessarily complex to nearly unreadable. Here's an excerpt from the first chapter:

The lack of interest in representing the daily grind of work routines in various forms of popular culture is perhaps understandable, as is the tendency among cultural critics to focus on the animation and meaningfulness of commodities rather than the eclipse of laboring activity that Marx identifies as the source of their fetishization (Marx 1976, 164-65). The preference for a level of abstraction that tends not to register either the qualitative dimensions or the hierarchical relations of work can also account for its relative neglect in the field of mainstream economics. But the lack of attention to the lived experiences and political textures of work within political theory would seem to be another matter. Indeed, political theorists tend to be more interested in our lives as citizens and noncitizens, legal subjects and bearers of rights, consumers and spectators, religious devotees and family members, than in our daily lives as workers.

This is only a quarter of a paragraph, and the entire book is written like this. I don't mind the occasional use of longer words for their precise meanings ("qualitative," "hierarchical") and can tolerate the academic habit of inserting mostly unnecessary citations. I have less patience with the meandering and complex sentences, excessive hedge words ("perhaps," "seem to be," "tend to be"), unnecessarily indirect phrasing ("can also account for" instead of "explains"), or obscure terms that are unnecessary to the sentence (what is "animation of commodities"?). And please have mercy and throw a reader some paragraph breaks. The writing style means substantial unnecessary effort for the reader, which is why it took me six months to read this book. It stalled all of my non-work non-fiction reading and I'm not sure it was worth the effort. That's unfortunate, because there were several important ideas in here that were new to me. The first was the overview of the "wages for housework" movement, which I had not previously heard of. It started from the common feminist position that traditional "women's work" is undervalued and advocated taking the next logical step of giving it equality with paid work by making it paid work. This was not successful, obviously, although the increasing prevalence of day care and cleaning services has made it partly true within certain economic classes in an odd and more capitalist way. While I, like Weeks, am dubious this was the right remedy, the observation that household work is essential to support capitalist activity but is unmeasured by GDP and often uncompensated both economically and socially has only become more accurate since the 1970s. Weeks argues that the usefulness of this movement should not be judged by its lack of success in achieving its demands, which leads to the second interesting point: the role of utopian demands in reframing and expanding a discussion. I normally judge a political demand on its effectiveness at convincing others to grant that demand, by which standard many activist campaigns (such as wages for housework) are unsuccessful. Weeks points out that making a utopian demand changes the way the person making the demand perceives the world, and this can have value even if the demand will never be granted. For example, to demand wages for housework requires rethinking how work is defined, what activities are compensated by the economic system, how such wages would be paid, and the implications for domestic social structures, among other things. That, in turn, helps in questioning assumptions and understanding more about how existing society sustains itself. Similarly, even if a utopian demand is never granted by society at large, forcing it to be rebutted can produce the same movement in thinking in others. In order to rebut a demand, one has to take it seriously and mount a defense of the premises that would allow one to rebut it. That can open a path to discussing and questioning those premises, which can have long-term persuasive power apart from the specific utopian demand. It's a similar concept as the Overton Window, but with more nuance: the idea isn't solely to move the perceived range of accepted discussion, but to force society to examine its assumptions and premises well enough to defend them, or possibly discover they're harder to defend than one might have thought. Weeks applies this principle to universal basic income, as a utopian demand that questions the premise that work should be central to personal identity. I kept thinking of the Black Lives Matter movement and the demand to abolish the police, which (at least in popular discussion) is a more recent example than this book but follows many of the same principles. The demand itself is unlikely to be met, but to rebut it requires defending the existence and nature of the police. That in turn leads to questions about the effectiveness of policing, such as clearance rates (which are far lower than one might have assumed). Many more examples came to mind. I've had that experience of discovering problems with my assumptions I'd never considered when debating others, but had not previously linked it with the merits of making demands that may be politically infeasible. The book closes with an interesting discussion of the types of utopias, starting from the closed utopia in the style of Thomas More in which the author sets up an ideal society. Weeks points out that this sort of utopia tends to collapse with the first impossibility or inconsistency the reader notices. The next step is utopias that acknowledge their own limitations and problems, which are more engaging (she cites Le Guin's The Dispossessed). More conditional than that is the utopian manifesto, which only addresses part of society. The least comprehensive and the most open is the utopian demand, such as wages for housework or universal basic income, which asks for a specific piece of utopia while intentionally leaving unspecified the rest of the society that could achieve it. The demand leaves room to maneuver; one can discuss possible improvements to society that would approach that utopian goal without committing to a single approach. I wish this book were better-written and easier to read, since as it stands I can't recommend it. There were large sections that I read but didn't have the mental energy to fully decipher or retain, such as the extended discussion of Ernst Bloch and Friedrich Nietzsche in the context of utopias. But that way of thinking about utopian demands and their merits for both the people making them and for those rebutting them, even if they're not politically feasible, will stick with me. Rating: 5 out of 10

Release 0.6.28 of the digest package arrived at CRAN earlier today, and has already been uploaded Debian as well. digest creates hash digests of arbitrary R objects (using the md5, sha-1, sha-256, sha-512, crc32, xxhash32, xxhash64, murmur32, spookyhash, and blake3 algorithms) permitting easy comparison of R language objects. It is a mature and widely-used as many tasks may involve caching of objects for which it provides convenient general-purpose hash key generation. This release comes eleven months after the previous releases and rounds out a number of corners. Continuous Integration was updated using r-ci. Several contribututors help with a small fix applied to avoid unaligned reads, a rewording for a help page as well as windows path encoding for in the vectorised use case. My CRANberries provides the usual summary of changes to the previous version. For questions or comments use the issue tracker off the GitHub repo. If you like this or other open-source work I do, you can now sponsor me at GitHub. For the first year, GitHub will match your contributions.

This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. Please report excessive re-aggregation in third-party for-profit settings.

prrd facilitates the parallel running [of] reverse dependency [checks] when preparing R packages. It is used extensively for Rcpp, RcppArmadillo, RcppEigen, BH, and others. The key idea of prrd is simple, and described in some more detail on its webpage and its GitHub repo. Reverse dependency checks are an important part of package development that is easily done in a (serial) loop. But these checks are also generally embarassingly parallel as there is no or little interdependency between them (besides maybe shared build depedencies). See the (dated) screenshot (running six parallel workers, arranged in a split byobu session). This release brings some new features I used of late when testing and re-testing reverse dependencies for Rcpp. Enqueuing jobs can now consider the most recent prior job queue file. This allows us to find new packages that were not part of the previous runs. We added a second toggle to also add those packages who failed in the previous run. Finally, the dequeue interface allows to specify a date (rather than defaulting to the current date, useful for long-running jobs or restarts). The release is summarised in the NEWS entry:

Changes in prrd version 0.0.5 (2021-09-22)

• Some remaing http URLs were changed to https.
• The dequeueJobs script has a new argument date to help specify a queue file.
• The enqueueJobs can now compute just a delta of (new) packages relative to a given prior queuefile and run.
• When running in delta mode, previously failed packages can also be selected.

My CRANberries provides the usual summary of changes to the previous version. See the aforementioned webpage and its repo for details. For more questions or comments use the issue tracker off the GitHub repo. If you like this or other open-source work I do, you can now sponsor me at GitHub.

This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. Please report excessive re-aggregation in third-party for-profit settings.

This post is about some changes recently made to Rust's ErrorKind, which aims to categorise OS errors in a portable way. Audiences for this post

• The educated general reader interested in a case study involving error handling, stability, API design, and/or Rust.
• Rust users who have tripped over these changes. If this is you, you can cut to the chase and skip to How to fix.

Background and context Error handling principles Handling different errors differently is often important (although, sadly, often neglected). For example, if a program tries to read its default configuration file, and gets a "file not found" error, it can proceed with its default configuration, knowing that the user hasn't provided a specific config. If it gets some other error, it should probably complain and quit, printing the message from the error (and the filename). Otherwise, if the network fileserver is down (say), the program might erroneously run with the default configuration and do something entirely wrong. Rust's portability aims The Rust programming language tries to make it straightforward to write portable code. Portable error handling is always a bit tricky. One of Rust's facilities in this area is std::io::ErrorKind which is an enum which tries to categorise (and, sometimes, enumerate) OS errors. The idea is that a program can check the error kind, and handle the error accordingly. That these ErrorKinds are part of the Rust standard library means that to get this right, you don't need to delve down and get the actual underlying operating system error number, and write separate code for each platform you want to support. You can check whether the error is ErrorKind::NotFound (or whatever). Because ErrorKind is so important in many Rust APIs, some code which isn't really doing an OS call can still have to provide an ErrorKind. For this purpose, Rust provides a special category ErrorKind::Other, which doesn't correspond to any particular OS error. Rust's stability aims and approach Another thing Rust tries to do is keep existing code working. More specifically, Rust tries to:

1. Avoid making changes which would contradict the previously-published documentation of Rust's language and features.
2. Tell you if you accidentally rely on properties which are not part of the published documentation.

By and large, this has been very successful. It means that if you write code now, and it compiles and runs cleanly, it is quite likely that it will continue work properly in the future, even as the language and ecosystem evolves. This blog post is about a case where Rust failed to do (2), above, and, sadly, it turned out that several people had accidentally relied on something the Rust project definitely intended to change. Furthermore, it was something which needed to change. And the new (corrected) way of using the API is not so obvious. Rust enums, as relevant to io::ErrorKind (Very briefly:) When you have a value which is an io::ErrorKind, you can compare it with specific values:

    if error.kind() == ErrorKind::NotFound   ...
  

But in Rust it's more usual to write something like this (which you can read like a switch statement):

    match error.kind()  
      ErrorKind::NotFound => use_default_configuration(),
      _ => panic!("could not read config file  :  ", &file, &error),
     
  

Here _ means "anything else". Rust insists that match statements are exhaustive, meaning that each one covers all the possibilities. So if you left out the line with the _, it wouldn't compile. Rust enums can also be marked non_exhaustive, which is a declaration by the API designer that they plan to add more kinds. This has been done for ErrorKind, so the _ is mandatory, even if you write out all the possibilities that exist right now: this ensures that if new ErrorKinds appear, they won't stop your code compiling. Improving the error categorisation The set of error categories stabilised in Rust 1.0 was too small. It missed many important kinds of error. This makes writing error-handling code awkward. In any case, we expect to add new error categories occasionally. I set about trying to improve this by proposing new ErrorKinds. This obviously needed considerable community review, which is why it took about 9 months. The trouble with Other and tests Rust has to assign an ErrorKind to every OS error, even ones it doesn't really know about. Until recently, it mapped all errors it didn't understand to ErrorKind::Other - reusing the category for "not an OS error at all". Serious people who write serious code like to have serious tests. In particular, testing error conditions is really important. For example, you might want to test your program's handling of disk full, to make sure it didn't crash, or corrupt files. You would set up some contraption that would simulate a full disk. And then, in your tests, you might check that the error was correct. But until very recently (still now, in Stable Rust), there was no ErrorKind::StorageFull. You would get ErrorKind::Other. If you were diligent you would dig out the OS error code (and check for ENOSPC on Unix, corresponding Windows errors, etc.). But that's tiresome. The more obvious thing to do is to check that the kind is Other. Obvious but wrong. ErrorKind is non_exhaustive, implying that more error kinds will appears, and, naturally, these would more finely categorise previously-Other OS errors. Unfortunately, the documentation note

Errors that are Other now may move to a different or a new ErrorKind variant in the future.

was only added in May 2020. So the wrongness of the "obvious" approach was, itself, not very obvious. And even with that docs note, there was no compiler warning or anything. The unfortunate result is that there is a body of code out there in the world which might break any time an error that was previously Other becomes properly categorised. Furthermore, there was nothing stopping new people writing new obvious-but-wrong code. Chosen solution: Uncategorized The Rust developers wanted an engineered safeguard against the bug of assuming that a particular error shows up as Other. They chose the following solution: There is now a new ErrorKind::Uncategorized which is now used for all OS errors for which there isn't a more specific categorisation. The fallback translation of unknown errors was changed from Other to Uncategorised. This is de jure justified by the fact that this enum has always been marked non_exhaustive. But in practice because this bug wasn't previously detected, there is such code in the wild. That code now breaks (usually, in the form of failing test cases). Usually when Rust starts to detect a particular programming error, it is reported as a new warning, which doesn't break anything. But that's not possible here, because this is a behavioural change. The new ErrorKind::Uncategorized is marked unstable. This makes it impossible to write code on Stable Rust which insists that an error comes out as Uncategorized. So, one cannot now write code that will break when new ErrorKinds are added. That's the intended effect. The downside is that this does break old code, and, worse, it is not as clear as it should be what the fixed code looks like. Alternatives considered and rejected by the Rust developers Not adding more ErrorKinds This was not tenable. The existing set is already too small, and error categorisation is in any case expected to improve over time. Just adding ErrorKinds as had been done before This would mean occasionally breaking test cases (or, possibly, production code) when an error that was previously Other becomes categorised. The broken code would have been "obvious", but de jure wrong, just as it is now, So this option amounts to expecting this broken code to continue to be written and continuing to break it occasionally. Somehow using Rust's Edition system The Rust language has a system to allow language evolution, where code declares its Edition (2015, 2018, 2021). Code from multiple editions can be combined, so that the ecosystem can upgrade gradually. It's not clear how this could be used for ErrorKind, though. Errors have to be passed between code with different editions. If those different editions had different categorisations, the resulting programs would have incoherent and broken error handling. Also some of the schemes for making this change would mean that new ErrorKinds could only be stabilised about once every 3 years, which is far too slow. How to fix code broken by this change Most main-line error handling code already has a fallback case for unknown errors. Simply replacing any occurrence of Other with _ is right. How to fix thorough tests The tricky problem is tests. Typically, a thorough test case wants to check that the error is "precisely as expected" (as far as the test can tell). Now that unknown errors come out as an unstable Uncategorized variant that's not so easy. If the test is expecting an error that is currently not categorised, you want to write code that says "if the error is any of the recognised kinds, call it a test failure". What does "any of the recognised kinds" mean here ? It doesn't meany any of the kinds recognised by the version of the Rust stdlib that is actually in use. That set might get bigger. When the test is compiled and run later, perhaps years later, the error in this test case might indeed be categorised. What you actually mean is "the error must not be any of the kinds which existed when the test was written". IMO therefore the right solution for such a test case is to cut and paste the current list of stable ErrorKinds into your code. This will seem wrong at first glance, because the list in your code and in Rust can get out of step. But when they do get out of step you want your version, not the stdlib's. So freezing the list at a point in time is precisely right. You probably only want to maintain one copy of this list, so put it somewhere central in your codebase's test support machinery. Periodically, you can update the list deliberately - and fix any resulting test failures. Unfortunately this approach is not suggested by the documentation. In theory you could work all this out yourself from first principles, given even the situation prior to May 2020, but it seems unlikely that many people have done so. In particular, cutting and pasting the list of recognised errors would seem very unnatural. Conclusions This was not an easy problem to solve well. I think Rust has done a plausible job given the various constraints, and the result is technically good. It is a shame that this change to make the error handling stability more correct caused the most trouble for the most careful people who write the most thorough tests. I also think the docs could be improved.

edited shortly after posting, and again 2021-09-22 16:11 UTC, to fix HTML slips

comments

Nearly 4 weeks after contracting COVID-19 I am finally able to return to work Yes I have had both Jabs (my 2nd dose was back in June), and this knocked me for six. I spent most of the time in bed, and only started to get up and about 10 days ago. I passed this on to both my wife and daughter (my wife has also been double jabbed), fortunately they didn t get it as bad as me and have been back at work / school for the last week. I also passed it on to a friend at the UK Debian BBQ, hosted once again by Sledge and Randombird, before I started showing symptoms. Fortunately (after a lot of PCR tests for attendees) it doesn t look like I passed it to anyone else I wouldn t wish this on anyone. I went on holiday back in August (still in England) thinking that having both jabs we would be fine. We stayed in self catering accommodation and we spent our time outside, we visited open air museums, walked around gardens etc, however we did eat out in relatively empty pubs and restaurants. And yes we did all have face masks on when we went indoors (although obviously we removed them whilst eating). I guess that is when I caught this, but have no idea exactly when or where. Even after vaccination, it is still possible to both catch and spread this virus. Fortunately having been vaccinated my resulting illness was (statistically) less bad than it would otherwise have been. I dread to think how bad I would have been if I had not already been vaccinated, I suspect that I would have ended up in ICU. I am still very tired, and have been told it may take many more weeks to get back to my former self. Other than being overweight, prior to this I have been in good health. If you are reading this and have NOT yet had a vaccine and one is available for you, please, please get it done.

In this blog post, I will cover a few related but also different topics around X2Go - the GNU/Linux based remote computing framework. Introduction and Catch Up For those, who haven't come across X2Go, so far... With X2Go [0] you can log into remote GNU/Linux machines graphically and launch headless desktop environments, seamless/published applications or access an already running desktop session (on a local Xserver or running as a headless X2Go desktop session) via X2Go's session shadowing / mirroring feature. Graphical backend: NXv3 For several years, there was only one graphical backend available in X2Go, the NXv3 software. In NXv3, you have a headless or nested (it can do both) Xserver that has some remote magic built-in and is able to transfer the Xserver's graphical data to a remote client (NX proxy). Over the wire, the NX protocol allows for data compression (JPEG, PNG, etc.) and combines it with bitmap caching, so that the overall result is a fast and responsive desktop experience even on low latency and low bandwidth connections. This especially applies to X desktop environments that use many native X protocol operations for drawing windows and widget onto the screen. The more bitmaps involved (e.g. in applications with client-side rendering of window controls and such), the worse the quality of a session experience. The current main maintainer of NVv3 (aka nx-libs [1]) is Ulrich Sibiller. Uli has my and the X2Go community's full appreciation, admiration and gratitude for all the work he does on nx-libs, constantly improving NXv3 without breaking compatibility with legacy use cases (yes, FreeNX is still alive, by the way). NEW: Alternative Graphical Backend: X2Go Kdrive Over the past 1.5 years, Oleksandr Shneyder (Alex), co-founder of X2Go, has been working on a re-implementation of an alternative, less X11-dependent graphical backend. The underlying Xserver technology is the kdrive part of the X.org server project. People on GNU/Linux might have used kdrive technology already: The Xephyr nested Xserver uses the kdrive implementation. The idea of the X2Go Kdrive [2] implementation in X2Go is providing a headless Xserver on the X2Go Server side for running X11 based desktop sessions inside while using an X11-agnostic data protocol for sending the graphical desktop data to the client-side for rendering. Whereas, with NXv3 technology, you need a local Xserver on the client side, with X2Go Kdrive you only need a client app(lication) that can draw bitmaps into some sort of framebuffer, such as a client-side X11 Xserver, a client-side Wayland compositor or (hold your breath) an HTMLv5 canvas in a web browser. X2Go Kdrive Client Implementations During first half of this year, I tested and DEB-packaged Alex's X2Go HTMLv5 client code [3] and it has been available for testing in the X2Go nightly builds archive for a while now. Of course, the native X2Go Client application has X2Go Kdrive support for a while, too, but it requires a Qt5 application in the background, the x2gokdriveclient (which is still only available in X2Go nightly builds or from X2Go Git [4]). X2Go and Remmina As currently posted by the Remmina community [5], one of my employees has been working on finalizing an already existing draft of mine for the last couple of months: Remmina Plugin X2Go. This project has been contracted by BAUR-ITCS UG (haftungsbeschr nkt) already a while back and has been financed via X2Go funding from one of their customers. Unfortunately, I never got around really to finalizing the project. Apologies for this. Daniel Teichmann, who has been in the company for a while now, but just recently switched to an employment model with considerably more work hours per week, now picked up this project two months ago and achieved awesome things on the way. Daniel Teichmann and Antenore Gatta (Remmina core developer, aka tmow) have been cooperating intensely on this, recently, with the objective of getting the X2Go plugin code merged into Remmina asap. We are pretty close to the first touchdown (i.e. code merge) of this endeavour. Thanks to Antenore for his support on this. This is much appreciated. Remmina Plugin X2Go - Current Challenges The X2Go Plugin for Remmina implementation uses Python X2Go (PyHoca-CLI) under the bonnet and basically does a system call to pyhoca-cli according to the session settings configured in the Remmina session profile UI. When using NXv3 based sessions, the session window appears on the client-side Xserver and immediately gets caught by Remmina and embedded into the Remmina frame (via Xembed protocol) where its remote sessions are supposed to appear. (Thanks that GtkSocket is still around in GTK-3). The knowing GTK-3 experts among you may have noticed: GtkSocket is obsolete and has been removed from GTK-4. Also, GtkSocket support is only available in GTK-3 when using its X11 rendering backend. For the X2Go Kdrive implementation, we tested a similar approach (embedding the x2gokdriveclient Qt5 window via Xembed/GtkSocket), but it seems that GtkSocket and Qt5 applications don't work well together and we did not succeed in embedding the Qt5 window of the x2gokdriveclient application into Remmina, so far. Also, this would be a work-around for the bigger problem: We want, long-term, provide X2Go Kdrive support in Remmina, not only for Remmina running with GTK-3/X11, but also when Remmina is used natively on top of Wayland. So, the more sustainable approach for showing an X2Go Kdrive based X2Go session in Remmina would be a GTK-3/4 or a Glib-2.0 + Cairo based rendering client provided as a shared library. This then could be used by Remmina for drawing the session bitmaps into the Remmina session frame. This would require a port of the x2gokdriveclient Qt code into a non-Qt implementation. However, we are running out of funding to make this happen at the moment. More Funding Needed for this Journey As you might guess, such a project as proposed is a project that some people do in their spare time, others do it for a living. I'd love to continue this project and have Daniel Teichmann continue his work on this, so that Remmina might soon be able to provide native X2Go Kdrive Client support. If people read this and are interested in supporting such a project, please get in touch [6]. Thanks so much! light+love
Mike (aka sunweaver) [0] https://wiki.x2go.org/
[1] https://github.com/ArcticaProject/nx-libs
[2] https://code.x2go.org/gitweb?p=x2gokdrive.git;a=tree
[3] https://code.x2go.org/gitweb?p=x2gohtmlclient.git;a=tree
[4] https://code.x2go.org/gitweb?p=x2gokdriveclient.git;a=tree
[5] https://remmina.org/x2go/
[6] https://das-netzwerkteam.de/

Another small release of the tidyCpp package arrived on CRAN overnight. The packages offers a clean C++ layer (as well as one small C++ helper class) on top of the C API for R which aims to make use of this robust (if awkward) C API a little easier and more consistent. See the vignette for motivating examples. The Protect class now uses the default methods for copy and move constructors and assignment allowing for wide use of the class. The small NumVec class now uses it for its data member. The NEWS entry (which I failed to update for the releases) follows.

Changes in tidyCpp version 0.0.5 (2021-09-16)

• The Protect class uses default copy and move assignments and constructors
• The data object in NumVec is now a Protect object

Thanks to my CRANberries, there is also a diffstat report for this release. For questions, suggestions, or issues please use the issue tracker at the GitHub repo. If you like this or other open-source work I do, you can now sponsor me at GitHub.

This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. Please report excessive re-aggregation in third-party for-profit settings.

Colson Whitehead's latest novel, Harlem Shuffle, was always going to be widely reviewed, if only because his last two books won Pulitzer prizes. Still, after enjoying both The Underground Railroad and The Nickel Boys, I was certainly going to read his next book, regardless of what the critics were saying indeed, it was actually quite agreeable to float above the manufactured energy of the book's launch. Saying that, I was encouraged to listen to an interview with the author by Ezra Klein. Now I had heard Whitehead speak once before when he accepted the Orwell Prize in 2020, and once again he came across as a pretty down-to-earth guy. Or if I were to emulate the detached and cynical tone Whitehead embodied in The Nickel Boys, after winning so many literary prizes in the past few years, he has clearly rehearsed how to respond to the cliched questions authors must be asked in every interview. With the obligatory throat-clearing of 'so, how did you get into writing?', for instance, Whitehead replies with his part of the catechism that 'It seemed like being a writer could be a cool job. You could work from home and not talk to people.' The response is the right combination of cute and self-effacing... and with its slight tone-deafness towards enforced isolation, it was no doubt honed before Covid-19. Harlem Shuffle tells three separate stories about Ray Carney, a furniture salesman and 'fence' for stolen goods in New York in the 1960s. Carney doesn't consider himself a genuine criminal though, and there's a certain logic to his relativistic morality. After all, everyone in New York City is on the take in some way, and if some 'lightly used items' in Carney's shop happened to have had 'previous owners', well, that's not quite his problem. 'Nothing solid in the city but the bedrock,' as one character dryly observes. Yet as Ezra pounces on in his NYT interview mentioned abov, the focus on the Harlem underworld means there are very few women in the book, and Whitehead's circular response ah well, it's a book about the criminals at that time! was a little unsatisfying. Not only did it feel uncharacteristically slippery of someone justly lauded for his unflinching power of observation (after all, it was the author who decided what to write about in the first place), it foreclosed on the opportunity to delve into why the heist and caper genres (from The Killing, The Feather Thief, Ocean's 11, etc.) have historically been a 'male' mode of storytelling. Perhaps knowing this to be the case, the conversation quickly steered towards Ray Carney's wife, Elizabeth, the only woman in the book who could be said possesses some plausible interiority. The following off-hand remark from Whitehead caught my attention:

My wife is convinced that [Elizabeth] knows everything about Carney's criminal life, and is sort of giving him a pass. And I'm not sure if that's true. I have to have to figure out exactly what she knows and when she knows it and how she feels about it.

I was quite taken by this, although not simply due to its effect on the story it self. As in, it immediately conjured up a charming picture of Whitehead's domestic arrangements: not only does Whitehead's wife feel free to disagree with what one of Whitehead's 'own' characters knows or believes, but that Colson has no problem whatsoever sharing that disagreement with the public at large. (It feels somehow natural that Whitehead's wife believes her counterpart knows more than she lets on, whilst Whitehead himself imbues the protagonist's wife with a kind of neo-Victorian innocence.) I'm minded to agree with Whitehead's partner myself, if only due to the passages where Elizabeth is studiously ignoring Carney's otherwise unexplained freak-outs. But all of these meta-thoughts simply underline just how emancipatory the Death of the Author can be. This product of academic literary criticism (the term was coined by Roland Barthes' 1967 essay of the same name) holds that the original author's intentions, ideas or biographical background carry no especial weight in determining how others should interpret their work. It is usually understood as meaning that a writer's own views are no more valid or 'correct' than the views held by someone else. (As an aside, I've found that most readers who encounter this concept for the first time have been reading books in this way since they were young. But the opposite is invariably true with cinephiles, who often have a bizarre obsession with researching or deciphering the 'true' interpretation of a film.) And with all that in mind, can you think of a more wry example of how freeing (and fun) nature of the Death of the Author than an author's own partner dissenting with their (Pulitzer Prize-winning) husband on the position of a lynchpin character?

The 1964 Harlem riot began after James Powell, a 15-year-old African American, was shot and killed by Thomas Gilligan, an NYPD police officer in front of 10s of witnesses. Gilligan was subsequently cleared by a grand jury.

As it turns out, the reviews for Harlem Shuffle have been almost universally positive, and after reading it in the two days after its release, I would certainly agree it is an above-average book. But it didn't quite take hold of me in the way that The Underground Railroad or The Nickel Boys did, especially the later chapters of The Nickel Boys that were set in contemporary New York and could thus make some (admittedly fairly explicit) connections from the 1960s to the present day that kind of connection is not there in Harlem Shuffle, or at least I did not pick up on it during my reading. I can see why one might take exception to that, though. For instance, it is certainly true that the week-long Harlem Riot forms a significant part of the plot, and some events in particular are entirely contingent on the ramifications of this momentous event. But it's difficult to argue the riot's impact are truly integral to the story, so not only is this uprising against police brutality almost regarded as a background event, any contemporary allusion to the murder of George Floyd is subsequently watered down. It's nowhere near the historical rubbernecking of Forrest Gump (1994), of course, but that's not a battle you should ever be fighting. Indeed, whilst a certain smoothness of affect is to be priced into the Whitehead reading experience, my initial overall reaction to Harlem Shuffle was fairly flat, despite all the action and intrigue on the page. The book perhaps belies its origins as a work conceived during quarantine after all, the book is essentially comprised of three loosely connected novellas, almost as if the unreality and mental turbulence of lockdown prevented the author from performing the psychological 'deep work' of producing a novel-length text with his usual depth of craft. A few other elements chimed with this being a 'lockdown novel' as well, particularly the book's preoccupation with the sheer physicality of the city compared to the usual complex interplay between its architecture and its inhabitants. This felt like it had been directly absorbed into the book from the author walking around his deserted city, and thus being able to take in details for the first time:

The doorways were entrances into different cities no, different entrances into one vast, secret city. Ever close, adjacent to all you know, just underneath. If you know where to look.

And I can't fail to mention that you can almost touch Whitehead's sublimated hunger to eat out again as well:

Stickups were chops they cook fast and hot, you re in and out. A stakeout was ribs fire down low, slow, taking your time. [ ] Sometimes when Carney jumped into the Hudson when he was a kid, some of that stuff got into his mouth. The Big Apple Diner served it up and called it coffee.

More seriously, however, the relatively thin personalities of minor characters then reminded me of the simulacrum of Zoom-based relationships, and the essentially unsatisfactory endings to the novellas felt reminiscent of lockdown pseudo-events that simply fizzle out without a bang. One of the stories ties up loose ends with: 'These things were usually enough to terminate a mob war, and they appeared to end the hostilities in this case as well.' They did? Well, okay, I guess.

The corner of 125th Street and Morningside Avenue in 2019, the purported location of Carney's fictional furniture store. Signage plays a prominent role in Harlem Shuffle, possibly due to the author's quarantine walks.

Still, it would be unfair to characterise myself as 'disappointed' with the novel, and none of this piece should be taken as really deep criticism. The book certainly was entertaining enough, and pretty funny in places as well:

Carney didn t have an etiquette book in front of him, but he was sure it was bad manners to sit on a man s safe. [ ] The manager of the laundromat was a scrawny man in a saggy undershirt painted with sweat stains. Launderer, heal thyself.

Yet I can't shake the feeling that every book you write is a book that you don't, and so we might need to hold out a little longer for Whitehead's 'George Floyd novel'. (Although it is for others to say how much of this sentiment is the expectations of a White Reader for The Black Author to ventriloquise the pain of 'their' community.) Some room for personal critique is surely permitted. I dearly missed the junk food energy of the dry and acerbic observations that run through Whitehead's previous work. At one point he had a good line on the model tokenisation that lurks behind 'The First Negro to...' labels, but the callbacks to this idea ceased without any payoff. Similar things happened with the not-so-subtle critiques of the American Dream:

Entrepreneur? Pepper said the last part like manure. That s just a hustler who pays taxes. [ ] One thing I ve learned in my job is that life is cheap, and when things start getting expensive, it gets cheaper still.

Ultimately, though, I think I just wanted more. I wanted a deeper exploration of how the real power in New York is not wielded by individual street hoodlums or even the cops but in the form of real estate, essentially serving as a synecdoche for Capital as a whole. (A recent take of this can be felt in Jed Rothstein's 2021 documentary, WeWork: Or the Making and Breaking of a $47 Billion Unicorn and it is perhaps pertinent to remember that the US President at the time this novel was written was affecting to be a real estate tycoon.). Indeed, just like the concluding scenes of J. J. Connolly's Layer Cake, although you can certainly pull off a cool heist against the Man, power ultimately resides in those who control the means of production... and a homespun furniture salesman on the corner of 125 & Morningside just ain't that. There are some nods to kind of analysis in the conclusion of the final story ('Their heist unwound as if it had never happened, and Van Wyck kept throwing up buildings.'), but, again, I would have simply liked more. And when I attempted then file this book away into the broader media landscape, given the current cultural visibility of 1960s pop culture (e.g. One Night in Miami (2020), Judas and the Black Messiah (2021), Summer of Soul (2021), etc.), Harlem Shuffle also seemed like a missed opportunity to critically analyse our (highly-qualified) longing for the civil rights era. I can certainly understand why we might look fondly on the cultural products from a period when politics was less alienated, when society was less atomised, and when it was still possible to imagine meaningful change, but in this dimension at least, Harlem Shuffle seems to merely contribute to this nostalgic escapism.

tl;dr dgit clone sourcepackage gets you the source code, as a git tree, in ./sourcepackage. cd into it and dpkg-buildpackage -uc -b. Do not use: "VCS" links on official Debian web pages like tracker.debian.org; "debcheckout"; searching Debian's gitlab (salsa.debian.org). These are good for Debian experts only. If you use Debian's "official" source git repo links you can easily build a package without Debian's patches applied.[1] This can even mean missing security patches. Or maybe it can't even be built in a normal way (or at all). OMG WTF BBQ, why? It's complicated. There is History. Debian's "most-official" centralised source repository is still the Debian Archive, which is a system based on tarballs and patches. I invented the Debian source package format in 1992/3 and it has been souped up since, but it's still tarballs and patches. This system is, of course, obsolete, now that we have modern version control systems, especially git. Maintainers of Debian packages have invented ways of using git anyway, of course. But this is not standardised. There is a bewildering array of approaches. The most common approach is to maintain git tree containing a pile of *.patch files, which are then often maintained using quilt. Yes, really, many Debian people are still using quilt, despite having git! There is machinery for converting this git tree containing a series of patches, to an "official" source package. If you don't use that machinery, and just build from git, nothing applies the patches. [1] This post was prompted by a conversation with a friend who had wanted to build a Debian package, and didn't know to use dgit. They had got the source from salsa via a link on tracker.d.o, and built .debs without Debian's patches. This not a theoretical unsoundness, but a very real practical risk. Future is not very bright In 2013 at the Debconf in Vaumarcus, Joey Hess, myself, and others, came up with a plan to try to improve this which we thought would be deployable. (Previous attempts had failed.) Crucially, this transition plan does not force change onto any of Debian's many packaging teams, nor onto people doing cross-package maintenance work. I worked on this for quite a while, and at a technical level it is a resounding success. Unfortunately there is a big limitation. At the current stage of the transition, to work at its best, this replacement scheme hopes that maintainers who update a package will use a new upload tool. The new tool fits into their existing Debian git packaging workflow and has some benefits, but it does make things more complicated rather than less (like any transition plan must, during the transitional phase). When maintainers don't use this new tool, the standardised git branch seen by users is a compatibility stub generated from the tarballs-and-patches. So it has the right contents, but useless history. The next step is to allow a maintainer to update a package without dealing with tarballs-and-patches at all. This would be massively more convenient for the maintainer, so an easy sell. And of course it links the tarballs-and-patches to the git history in a proper machine-readable way. We held a "git packaging requirements-gathering session" at the Curitiba Debconf in 2019. I think the DPL's intent was to try to get input into the git workflow design problem. The session was a great success: my existing design was able to meet nearly everyone's needs and wants. The room was obviously keen to see progress. The next stage was to deploy tag2upload. I spoke to various key people at the Debconf and afterwards in 2019 and the code has been basically ready since then. Unfortunately, deployment of tag2upload is mired in politics. It was blocked by a key team because of unfounded security concerns; positive opinions from independent security experts within Debian were disregarded. Of course it is always hard to get a team to agree to something when it's part of a transition plan which treats their systems as an obsolete setup retained for compatibility. Current status If you don't know about Debian's git packaging practices (eg, you have no idea what "patches-unapplied packaging branch without .pc directory" means), and don't want want to learn about them, you must use dgit to obtain the source of Debian packages. There is a lot more information and detailed instructions in dgit-user(7). Hopefully either the maintainer did the best thing, or, if they didn't, you won't need to inspect the history. If you are a Debian maintainer, you should use dgit push-source to do your uploads. This will make sure that users of dgit will see a reasonable git history.

edited 2021-09-15 14:48 Z to fix a typo

comments

This is the fifth and final post in a series about the interface description language Candid. If you made it this far, you now have a good understanding of what Candid is, what it is for and how it is used. For this final post, I ll put the spotlight on specific aspects of Candid that are maybe surprising, or odd, or quirky. This section will be quite opinionated, and could maybe be called what I d do differently if I d re-do the whole thing . Note that these quirks are not serious problems, and they don t invalidate the overall design. I am writing this up not to discourage the use of Candid, but merely help interested parties to understand it better.

References in the wire format When the work on Candid began at DFINITY, the Internet Computer was still far away from being a thing, and many fundamental aspects about it were still in the air. I particular, there was still talk about embracing capabilities as a core feature of the application model, which would be implemented as opaque references on the system level, likely building on WebAssembly s host reference type proposal (which only landed recently), and could be used to model access permissions, custom tokens and many other things. So Candid is designed with that in mind, and you ll find that its wire format is not just a type table and a value table, but actually

a triple (T, M, R), where T ( type ) and M ( memory ) are sequences of bytes and R ( references ) is a sequence of references.

Also the wire format for values of function service tyeps have an extra byte to distinguish between public references (represented by a principal and possible a method name in the data part), and these opaque references. Alas, references never made it into the Internet Computer, so all Candid implementations simply ignore that part of the specification. But it s still in the spec, and if it confused you before, now you know why.

Hashed field names Candid record and variant types look like they have textual field names:

type T = record   syndactyle : nat; trustbuster: bool  

But that is actually only true superficially. The wire format for Candid only stores hashes of field names. So the above is actually equivalent to

type T = record   4260381820 : nat; 3504418361 : bool  

or, for that matter, to

type T = record   destroys : bool; rectum : nat  

(Yes, I used an english word list to find these hash collisions. There aren t that many actually.) The benefit of such hashing is that the messages are a bit smaller in most (not all) cases, but it is a big annoyance for dynamic uses of Candid. It s the reason why tools like dfx, if they don t know the Candid interface of a service, will print the result with just the numerical hash, letting you guess which field is which. It also complicates host languages that derive Candid types from the host language, like Motoko, as some records (e.g. record trustbuster: bool; destroys : int ) with field name hash collisions can not be represented in Candid, and either the host language s type system needs to be Candid aware now (as is the case of Motoko), or serialization/deserialization will fail at runtime, or odd bugs can happen. (More discussion of this issue).

Tuples Many languages have a built-in notion of a tuple type (e.g. (Int, Bool)), but Candid does not have such a type. The only first class product type is records. This means that tuples have to encoded as records somehow. Conveniently(?) record fields are just numbers after all, so the type (Int, Bool) would be mapped to the type

record   0 : int; 1 : bool  

So tuples can be expressed. But from my experience implementing the type mappings for Motoko and Haskell this is causing headaches. To get a good experience when importing from Candid, the tools have to try to reconstruct which records may have been tuples originally, and turn them into tuples. The main argument for the status quo is that Candid types should be canonical, and there should not be more than one product type, and records are fine, and there needs to be no anonymous product type. But that has never quite resonated with me, given the practical reality of tuple types in many host languages.

Argument sequences Did I say that Candid does not have tuple types? Turns out it does, sort of. There is no first class anonymous product, but since functions take sequences of arguments and results, there is a tuple type right there:

func foo : (bool, int) -> (int, bool)

Again, I found that ergonomic interaction with host languages becomes relatively unwieldy by requiring functions to take and return sequences of values. This is especially true for languages where functions take one argument value or return one result type (the latter being very common). Here, return sequences of length one are turned into that type directly, longer argument sequences turn into the host language s tuple type, and nullary argument sequences turn into the idiomatic unit type. But this means that the types (int, bool) -> () and (record 0: int, 1: bool ) -> () may be mapped to the same host language type, which causes problems when you hope to encode all necessary Candid type information in the host language. Another oddity with argument and result sequences is that you can give names to the entries, e.g. write

func hello : (last_name : text; first_name : text) -> ()

but these names are completely ignored! So while this looks like you can, for example, add new optional arguments in the middle, such as

func hello : (last_name : text; middle_name: opt text, first_name : text) -> ()

without breaking clients, this does not have the effect you think it has and will likely break. My suggestion is to never put names on function arguments and result values in Candid interfaces, and for anything that might be extended with new fields or where you want to name the arguments, use a single record type as the only argument:

func hello : (record   last_name : text; first_name : text ) -> ()

This allows you to add and remove arguments more easily and reliably.

Type shorthands The Candid specification defines a system of types, and then adds a number of syntactic short-hands . For example, if you write blob in a Candid type description, it ought to means the same as vec nat8. My qualm with that is that it doesn t always mean the same. A Candid type description is interpreted by a number of, say, consumers . Two such consumers are part of the Candid specification:

• The specification that defines the wire format for that type
• The upgrading (subtyping) rules

But there are more! For every host language, there is some mapping from Candid types to host language types, and also generic tools like Candid UI are consumers of the type algebra. If these were to take the Candid specification as gospel, they would be forced to treat blob and vec nat8 the same, but that would be quite unergonomic and might cause performance regressions (most language try to map blob to some compact binary data type, while vec t tends to turn into some form of array structure). So they need to be pragmatic and treat blob and vec nat8 differently. But then, for all practical purposes, blob is not just a short-hand of vec nat8. They are different types that just happens to have the same wire representations and subtyping relations. This affects not just blob, but also tuples (record blob; int; bool ) and field names , as discussed above.

The value text format For a while after defining Candid, the only implementation was in Motoko, and all the plumbing was automatic, so there was never a need for users to to explicitly handle Candid values, as all values were Motoko values. Still, for debugging and testing and such things, we eventually needed a way to print out Candid values, so the text format was defined ( To enable convenient debugging, the following grammar specifies a text format for values ). But soon the dfx tool learned to talk to canisters, and now users needed to enter Candid value on the command line, possibly even when talking to canisters for which the interface was not known to dfx. And, sure enough, the textual interface defined in the Candid spec was used. Unfortunately, since it was not designed for that use case, it is rather unwieldy:

• It is quite verbose. You have to write record , not just . Vectors are written vec ; instead of some conventional syntax like [ , ]. Variants are written as variant error = " " with braces that don t any value here, and something like #error " " might have worked as well. With a bit more care, a more concise and ergonomic syntax might have been possible.
• It wasn t designed to be sufficient to create a Candid value from it. If you write 5 it s unclear whether that s a nat or an int16 or what (and all of these have different wire representations). Type annotations were later added, but are relatively unwieldy, and don t cover all cases (e.g. a service reference with a recursive type cannot be represented in the textual format at the moment).
• Not really the fault of the textual format, but some useful information about the types is not reflected in the type description that s part of the wire format. In particular not the field names, and whether a value was intended to be binary data (blob) or a list of small numbers (vec nat8), so pretty-printing such values requires guesswork. The Haskell library even tries to brute-force the hash to guess the field name, if it is short or in a english word list!

In hindsight I think it was too optimistic to assume that correct static type information is always available, and instead of actively trying to discourage dynamic use, Candid might be better if we had taken these (unavoidable?) use cases into account.

Custom wire format At the beginning of this post, I have a Candid is list. The list is relatively long, and the actual wire format is just one bullet point. Yes, defining a wire format that works isn t rocket science, and it was easiest to just make one up. But since most of the interesting meat of Candid is in other aspects (subtyping rules, host language integration), I wonder if it would have been better to use an existing, generic wire format, such as CBOR, and build Candid as a layer on top. This would give us plenty of tools and libraries to begin with. And maybe it would have reduced barrier of entry for developers, which now led to the very strange situation that DFINITY advocates for the universal use of Candid on the Internet Computer, so that all services can smoothly interact, but two of the most important services on the Internet Computer (the registry and the ledger) use Protobuf as their primary interface format, with Candid interfaces missing or an afterthought.

Sideways Interface Evolution This is not a quirk of Candid itself, but rather an idiom of how you can use Candid that emerged from our solution for record extensions and upgrades. Consider our example from before, a service with interface

service   add_user : (record   login : text; name : text  ) -> ()  

where you want to add an age field, which should be a number. The official way of doing that is to add that field with an optional type:

service   add_user : (record   login : text; name : text; age : opt nat  ) -> ()  

As explained above, this will not break old clients, as the decoder will treat a missing argument as null. So far so good. But often when adding such a field you don t want to bother new clients with the fact that this age was, at some point in the past, not there yet. And you can do that! The trick is to distinguish between the interface you publish and the interface you implement. You can (for example in your documentation) state that the interface is

service   add_user : (record   login : text; name : text; age : nat  ) -> ()  

which is not a subtype of the old type, but it is the interface you want new clients to work with. And then your implementation uses the type with opt nat. Calls from old clients will come through as null, and calls from new clients will come through as opt 42. We can see this idiom used in the Management Canister of the Internet Computer. The current documented interface only mentions a controllers : vec principal field in the settings, but the implementation still can handle both the old controller : principal and the new controllers field. It s probably advisable to let your CI system check that new versions of your service continue to implement all published interfaces, including past interfaces. But as long as the actual implementation s interface is a subtype of all interfaces ever published, this works fine. This pattern is related to when your service implements, say, http_request (so its implemented interface is a subtype of that common interface), but does not include that method in the published documentation (because clients of your service don t need to call it).

Self-describing Services As you have noticed, Candid was very much designed assuming that all parties always have the service type of services they want to interact with. But the Candid specification does not define how one can obtain the interface of a given service, and there isn t really a an official way to do that on the Internet Computer. That is unfortunate, because many interesting features depend on that: Such as writing import C "ic:7e6iv-biaaa-aaaaf-aaada-cai" in your Motoko program, and having it s type right there. Or tools like ic.rocks, that allow you to interact with any canister right there. One reason why we don t really have that feature yet is because of disagreements about how dynamic that feature should be. Should you be able to just ask the canister for its interface (and allow the canister to vary the response, for example if it can change its functionality over time, even without changing the actual wasm code)? Or is the interface a static property of the code, and one should be able to query the system for that data, without the canister s active involvement. Or, completely different, should interfaces be distributed out of band, maybe together with API documentation, or in some canister registry somewhere else? I always leaned towards the first of these options, but not convincingly enough. The second options requires system assistance, so more components to change, more teams to be involved that maybe intrinsically don t care a lot about this feature. And the third might have emerged as the community matures and builds such infrastructure, but that did not happen yet. In the end I sneaked in an implementation of the first into Motoko, arguing that even if we don t know yet how this feature will be implemented eventually, we all want the feature to exist somehow, and we really really want to unblock all the interesting applications it enables (e.g. Candid UI). That s why every Motoko canister, and some rust canisters too, implements a method

__get_candid_interface_tmp_hack : () -> (text)

that one can use to get the Candid interface file. The name was chosen to signal that this may not be the final interface, but like all good provisional solutions, it may last longer than intended. If that s the case, I m not sorry. This concludes my blog post series about Candid, for now. If you want to know more, feel free to post your question on the DFINTY developer forum, and I ll probably answer.

This release is the culmination of a project that I've been wanting to do for two years, but nearly all the work was done in the past week. That experience made me rethink some of my skepticism, but I'll get to that part of the story later. In March of 1999, I got tired of writing HTML by hand and wrote a small program called spin that implemented a macro language that translated into HTML. This makes it one of the oldest programs for which I have a continuous development history, predating podlators by three months. I think only News::Gateway (now very dormant) and Term::ANSIColor (still under active development but very stable) are older, as long as I'm not counting orphaned packages like newsyslog. I've used spin continuously ever since. It's grown features and an ecosystem of somewhat hackish scripts to do web publishing things I've wanted over the years: journal entries like this one, book reviews, a simple gallery (with some now-unfortunate decisions about maximum image size), RSS feeds, and translation of lots of different input files into HTML. But the core program itself, in all those years, has been one single Perl script written mostly in my Perl coding style from the early 2000s before I read Perl Best Practices. My web site is long overdue for an overhaul. Just to name a couple of obvious problems, it looks like trash on mobile browsers, and I'm using URL syntax from the early days of the web that, while it prompts some nostalgia for tildes, means all the URLs are annoyingly long and embed useless information such as the fact each page is written in HTML. Its internals also use a lot of ad hoc microformats (a bit of RFC 2822 here, a text-based format with significant indentation there, a weird space-separated database) and are supported by programs that extract meaning from human-written pages and perform automated updates to them rather than having a clear separation between structure and data. This will be a very large project, but it's the sort of quixotic personal project that I enjoy. Maintaining my own idiosyncratic static site generator is almost certainly not an efficient use of my time compared to, say, converting everything to Hugo. But I have 3,428 pages (currently) written in the thread macro language, plus numerous customizations that cater to my personal taste and interests, and, most importantly, I like having a highly customized system that I know exactly how to automate. The blocker has been that I didn't want to work on spin as it existed. It badly needed a structural overhaul and modernization, and even more badly needed a test suite since every release involved tedious manual testing by pouring over diffs between generations of the web site. And that was enough work to be intimidating, so I kept putting it off. I've separately been vaguely aware that I have been spending too much time reading Twitter (specifically) and the news (in general). It would be one thing if I were taking in that information to do something productive about it, but I haven't been. It's just doomscrolling. I've been thinking about taking a break for a while but it kept not sticking, so I decided to make a concerted effort this week. It took about four days to stop wanting to check Twitter and forcing myself to go do something else productive or at least play a game instead. Then I managed to get started on my giant refactoring project, and holy shit, Twitter has been bad for my attention span! I haven't been able to sustain this level of concentration for hours at a time in years. Twitter's not the only thing to blame (there are a few other stressers that I've fixed in the past couple of years), but it's obviously a huge part. Anyway, this long personal ramble is prelude to the first release of DocKnot that includes my static site generator. This is not yet the full tooling from my old web tools page; specifically, it's missing faq2html, cl2xhtml, and cvs2xhtml. (faq2html will get similar modernization treatment, cvs2xhtml will probably be rewritten in Perl since I have some old, obsolete scripts that may live in CVS forever, and I may retire cl2xhtml since I've stopped using the GNU ChangeLog format entirely.) But DocKnot now contains the core of my site generation system, including the thread macro language, POD conversion (by way of Pod::Thread), and RSS feeds. Will anyone else ever use this? I have no idea; realistically, probably not. If you were starting from scratch, I'm sure you'd be better off with one of the larger and more mature static site generators that's not the idiosyncratic personal project of one individual. It is packaged for Debian because it's part of the tool chain for generating files (specifically README.md) that are included in every package I maintain, and thus is part of the transitive closure of Debian main, but I'm not sure anyone will install it from there for any other purpose. But for once making something for someone else isn't the point. This is my quirky, individual way to maintain web sites that originated in an older era of the web and that I plan to keep up-to-date (I'm long overdue to figure out what they did to HTML after abandoning the XHTML approach) because it brings me joy to do things this way. In addition to adding the static site generator, this release also has the regular sorts of bug fixes and minor improvements: better formatting of software pages for software that's packaged for Debian, not assuming every package has a TODO file, and ignoring Autoconf 2.71 backup files when generating distribution tarballs. You can get the latest version of DocKnot from CPAN as App-DocKnot, or from its distribution page. I know I haven't yet updated my web tools page to reflect this move, or changed the URL in the footer of all of my pages. This transition will be a process over the next few months and will probably prompt several more minor releases.

Debian 11 (Bullseye) comes with the mlt framework 6.24.0 and kdenlive 20.12.3. Unfortunately it was already too late (freeze) to build mlt with enabled OpenCV features, this are also required to use the motion tracker features, which are now still missing in the pure stable release. So I have just uploaded, migrated and backported mlt 6.26.1 along with kdenlive 21.04.3 to bullseye-backports with opencv support. :-) If you want to install it, just add the backports repository and install the new version with: apt install kdenlive/bullseye-backports

A week after the 0.2.5 release bringing the recent Google Summer of Code for RcppSMC to CRAN, we have a minor bug-fix release consistently, essentially, of one line. Everybody s favourite OS and toolchain did not know what to make of pow(), and I seemingly failed to test there, so shame on me. But now all is good thanks to proper use of std::pow(). RcppSMC provides Rcpp-based bindings to R for the Sequential Monte Carlo Template Classes (SMCTC) by Adam Johansen described in his JSS article. Sequential Monte Carlo is also referred to as Particle Filter in some contexts. The package now features the Google Summer of Code work by Leah South in 2017, and by Ilya Zarubin in 2021. This release is summarized below.

Changes in RcppSMC version 0.2.5 (2021-09-09)

• Compilation under Solaris is aided via std::pow use (Dirk in #65 fixing #64)

Courtesy of my CRANberries, there is a diffstat report for this release. More information is on the RcppSMC page. Issues and bugreports should go to the GitHub issue tracker. If you like this or other open-source work I do, you can now sponsor me at GitHub.

This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. Please report excessive re-aggregation in third-party for-profit settings.

The RcppSimdJson team is happy to share that a new version 0.1.6 arrived on CRAN earlier today. Its release coincides with release 1.0.0 of simdjson itself, which is included in this release too! RcppSimdJson wraps the fantastic and genuinely impressive simdjson library by Daniel Lemire and collaborators. Via very clever algorithmic engineering to obtain largely branch-free code, coupled with modern C++ and newer compiler instructions, it results in parsing gigabytes of JSON parsed per second which is quite mindboggling. The best-case performance is faster than CPU speed as use of parallel SIMD instructions and careful branch avoidance can lead to less than one cpu cycle per byte parsed; see the video of the talk by Daniel Lemire at QCon (also voted best talk). This version brings the new upstream release, thanks to a comprehensive pull request by Daniel Lemire. The short NEWS entry follows.

Changes in version 0.1.6 (2021-09-07)

• The C++17 dependency was stated more clearly in the DESCRIPTION file (Dirk)
• The simdjson version was updated to release 1.0.0 (Daniel Lemire in #70)

We should point out that the package still has a dependency on C++17 even though simdjson no longer does. Some of our earlier wrapping code uses it, this could be changed. If you, dear reader, would like to work on this please get in touch. Courtesy of my CRANberries, there is also a diffstat report for this release. For questions, suggestions, or issues please use the issue tracker at the GitHub repo. If you like this or other open-source work I do, you can now sponsor me at GitHub.

This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. Please report excessive re-aggregation in third-party for-profit settings.

This is the third post in a series about the interface description language Candid.

The record extension problem Initially, the upgrade rules of Candid were a straight-forward application of the canonical subtyping rules. This worked and was sound, but it forbid one very commonly requested use case: Extending records in argument position. The subtyping rule for records say that

   record   old_field : nat; new_field : int  
<: record   old_field : nat  

or, in words, that subtypes can have more field than supertypes. This makes intuitive sense: A user of a such a record value that only expects old_field to be there is not bothered if suddenly a new_field appears. But a user who expects new_field to be there is thrown off if it suddenly isn t anymore. Therefore it is ok to extend the records returned by a service with new fields, but not to extend the record in your methods s argument types. But users want to extend their record types over time, also in argument position! In fact, they often want to extend them in both positions. Consider a service with the following interface, where the CUser record appears both in argument and result position:

type CUser = record   login : text; name : text  ;
service C :  
  register_user : (CUser) -> ();
  get_user_data : (text) -> (CUser);
 

It seems quite natural to want to extend the record with a new field (say, the age of the user). But simply changing the definition of CUser to

type CUser = record   login : text; name : text; age : nat  

is not ok, because now register_user requires the age field, but old clients don t provide it. So how can we allow developers to make changes like this (because they really really want that), while keeping the soundness guarantees made by Candid (because we really really want that)? This question has bothered us for over two years, and we even created a meta issue that records the dozen approached we considered, discussed and eventually ditched.

Dynamic subtyping checks in opt I will spare you the history lesson, though, and explain the solution we eventually came up with. In the example above it seems unreasonable to let the developer add a field age of type nat. Since there may be old clients around, the service unavoidably has to deal with records that don t have an age field. If the code expects an age value, what should it do in that case? But one can argue that changing the record type to

type CUser = record   login : text; name : text; age : opt nat  

could work: If the age field is present, use the value. If the field is absent, inject a null value during decoding. In the first-order case, this rather obvious solution would work just fine, and we d be done. But Candid aims to solve the higher order case, and I said things get tricky here, didn t I? Consider another, independent service D with the following interface:

type DUser = record   login : text; name : text  ;
service D :  
  on_user_added : (func (DUser) -> ()) -> ()
 

This service has a method that takes a method reference, presumably with the intention of calling it whenever a new user was added to service D. And because the types line up nicely, we can compose these two services, by once calling D.on_user_added(C.register_user), so that from now on D calls C.register_user( ). These kind of service mesh-ups are central to the vision of the Internet Computer! But what happens if the services now evolve their types in different ways, e.g. towards

type CUser = record   login : text; name : text; age : opt nat  

and

type DUser = record   login : text; name : text; age : opt variant   under_age; adult  

Individually, these are type changes that we want to allow. But now the call from D to C may transmit a value of record ; age = opt variant under_age when C expects a natural number! This is precisely what we want to prevent by introducing types, and it seems we have lost soundness. The best solution we could find is to make the opt type somewhat special, and apply these extra rules when decoding at an expected type of opt t.

• If this was a record field, and the provided record value doesn t even have a field of that name, don t fail but instead treat this as null. This handles the first-order example above.
• If there is a value, look at its type (which, in Candid, is part of the message).
  • If it is opt t' and t' <: t, then decode as normal.
  • If it is opt t' but t' <: t does not hold, then treat that as null. This should only happen in these relatively obscure higher-order cases where services evolved differently and incompatibly, and makes sure that the calls that worked before the upgrades will continue to work afterwards. It is not advisable to actually use that rule when changing your service s interface. Tools that assist the developer with an upgrade check should prevent or warn about the use of this rule.
• Not strictly required here, but since we are making opt special anyways: If its type t' is not of the form opt , pretend it was opt t', and also pretend that the given value was wrapped in opt. This allows services to make record field in arguments that were required in the old version optional in a new version, e.g. as a way to deprecated them. So it is mildly useful, although I can report that it makes the meta-theory and implementation rather complex, in particular together with equirecursion and beasts like type O = opt O. See this discussion for a glimpse of that.

In the above I stress that we look at the type of the provided value, and not just the value. For example, if the sender sends the value opt vec at type opt vec nat, and the recipient expects opt vec bool, then this will decode as null, even though one could argue that the value opt vec could easily be understood at type opt vec bool. We initially had that, but then noticed that this still breaks soundness when there are references inside, and we have to do a full subtyping check in the decoder. This is very unfortunate, because it makes writing a Candid decoder a noticeably harder task that requires complicated graph algorithms with memoization (which I must admit Motoko has not implemented yet), but it s the least bad solution we could find so far.

Isn t that just dynamic typing? You might have noticed that with these rules, t <: opt t' holds for all types t and t'. In other words, every opt is a top type (like reserved), thus all optional types are equivalent. One could argue that we are throwing all of the benefits of static typing over board this way. But it s not quite as bad: It s true that decoding Candid values now involves a dynamic check inserting null values in certain situations, but as long as everyone plays by the rules (i.e. upgrades their services according to the Candid safe upgrade rules, and heeds the warning alluded to above), these certain situations will only happen in the obscurest of cases. It is, by the way, not material to this solution that the special subtyping behavior is applied to the opt type, and a neighboring point in the design space would be a dedicated upgraded t type constructor. That would allow developers to use the canonical opt type with the usual subtyping rules and communicate their intent ( clients should consider this field optional vs. old clients don t know about this field, but new clients should use it ) more cleanly at the expense of having more non-canonical types in the type system. To see why it is convenient if Candid has mostly just the normal types, read the next post, which will describe how Candid can be integrated into a host language.