Anuradha Weeraman: Because We Can


Following up on my previous post, I am releasing
debvulns-exporter, a Prometheus exporter
for tracking Debian system vulnerabilities. The underlying vulnerability
analysis logic remains identical to the previously released MCP server and CLI
utility.
Rather than constructing the Grafana dashboard from scratch, I used Claude 4.6
Sonnet via Antigravity to generate the layout configuration. The generated
dashboard for the local testbed functions effectively:
The complete, ready-to-import Grafana dashboard configuration is included
directly in the debvulns source code.
(no, this isn't a blog post about Joy Division songs)
Last time I wrote about Interzone, I was discussing issue #294, the
first published under new management in a paperback-sized format
("JB6"). The format and presentation of the magazine was fantastic: it fit in a
lot of my pockets, and was packed with 15 stories as well as the regular
columns, in full colour with fantastic layouts and illustrations. Sadly there
was only one more physical issue before Interzone was forced to become a
digital-only publication.
I don't want to dwell on the sad necessity to move to digital. Interzone
continues on, celebrating the milestone issue #300 in 2024. Subscriptions
are managed via Patreon.
Issue #305 just came out.
Instead I wanted to write a small bit about how I engaged with the paper
magazine, and the difficulties I've had trying to engage with not just
Interzone but any magazine-style publication in a digital context.
With most fiction, I read linearly: start the beginning and read to the end, in
order. That works well for me with e-readers. But for magazines (and
most non-fiction) I don't, I jump around: usually starting with the
table of contents, I might pick a short column to start, or jump into the
middle of the "book reviews" section to read about a specific book. I might
skip sections entirely. I find it very difficult to read like this with an
e-reader. I think this is partly because I reference the depth of the
paper book or magazine, its thickness, to orient myself. But it's also partly
the limitations of e-ink.
For print-Interzone, I used to start by inserting a small piece of paper
inside the cover (the delivery slip was ideal). On this I listed the stories
within and ticked them off when I read them (sometimes I double-ticked if I
really liked a story). That helped me to remember, perhaps months or years
later, whether I'd read all the stories or not, and which I liked.
I could do something similar on some e-readers: the Remarkable for
instance. But it's far from convenient to do on most e-ink devices.
Interzone digital is available as both ePUB, the most common format for e-books, and PDF. For reading on my
regular Kobo e-reader, PDFs don't work very well at all. I think this is
generally true of most e-readers.
Interzone was (and is) a well-designed magazine. The value of it was not
just the content of the text, but the context: how the stories were
presented; the accompanying art (most often colour in recent decades),
but also the typesetting. ePUB
doesn't specify much of that stuff
exactly: it leaves that up to the client and the client's preferences.
And there's a lot of advantages to that:
Prefer a different font face or size? No problem. And
most importantly for accessibility:
If reading in ePUB makes Interzone available to more readers then that's
a great thing. But sadly a lot is lost, IMHO.
The solution I'm trying is to read the PDF version on the iPad Mini I
resurrected earlier in the year.
Despite being an Internet tablet, since it's not really usable for browsing the
web anymore it's strangely still a distraction-free device. In fact it's pretty
much single-purpose for reading Interzone and the odd other book which benefits
from being read as PDF. I can appreciate the stylistic choices made in the
page-setting as they were intended; I can quickly jump around the issue without
waiting for an e-ink refresh; I get full colour; and whilst it would be tiring
to read for a long time on the iPad screen, for the length of articles or
stories in a magazine, this isn't a problem.
It's not a solution for tracking what I've read (that version of ipadOS is too
old to support clumsily scrawling on PDF pages with your fingers, at least in
the Books app) but it otherwise seems to work well, so I'll see how it goes.
.IP and
.TP macros and misrenders .IP by removing all space between
the tag and the text. Ideally groff bugs should be fixed in groff, but
apparently this rendering bug was
introduced intentionally by the groff maintainer to force authors who
were using .IP with text tags to switch to .TP for correct
formatting, allowing future introduction of a semantic distinction between
the two macros. I didn't see a good alternative at this relatively late
date after the release other than changing Pod::Man accordingly.
This will at least work around this problem for Pod::Man users,
although it won't help with existing manual pages.
This release also works around another backwards-incompatible change to
groff that attempts to force the default enabling of hyphenation and full
justification after every occurrence of the .TH macro. The groff
upstream position is currently that the end user should be able to set
registers and strings to override the defaults of hyphenation and full
justification, but the man page author has no control over the defaults
for these settings. Pod::Man ignores the admonishment in groff_man(7) and
overrides these registers anyway to restore its long-standing historic
behavior of always using left justification and disabling hyphenation,
because there is currently no way to change the default without overriding
the new user preference. Should some mechanism be provided in the future,
I'll be happy to adopt it and thus honor user configuration as well.
New in this release is support for an encoding of none, which tells
Pod::Man and Pod::Text to do no character set encoding in their output and
leave the output in Perl's internal representation. This is useful in
combination with output_string() when the output will be used
internally by a Perl program.
Pod::Man also adopts CR as the default fixed-width font instead
of its long-standing default of CW, originally chosen for
compatibility with Solaris. This avoids warnings with newer groff at the
cost of breaking troff (not nroff) output on Solaris 10. I believe this
platform is now sufficiently old, and this use case sufficiently obscure,
that no one will miss it. Solaris 11 and later will render man pages
correctly with troff, and --fixed=CW will restore the previous
behavior.
This release also has a few other bug fixes, particularly for quoting
heuristics in C<> blocks, and various documentation improvements.
You can get the latest release from CPAN or from
the podlators distribution page.
flatpak install flathub dev.zed.Zed
In Zed: open the Extensions view and install PHP.
Configure the debugger:
Create ~/.var/app/dev.zed.Zed/config/zed/debug.json:
[
"label": "PHP: Listen to Xdebug",
"adapter": "Xdebug",
"request": "launch",
"port": 9003,
"pathMappings":
"/var/www/html": "/home/<user>/devel/nextcloud/server",
"/var/www/html/apps-extra": "/home/<user>/devel/nextcloud/server/apps-extra",
"/var/www/html/apps-shared": "/home/<user>/devel/nextcloud/apps-shared"
]
Add one entry per bind-mounted app directory.
After creating the file, restart Zed.
Inside Zed, select "debugger: start" from command palette and then "PHP: Listen to Xdebug".
Verify Zed is listening. Running ss -tlnp grep 9003 on the host should show *:9003 with Zed as the process.
/usr/local/etc/php/conf.d/xdebug.ini:
xdebug.mode = debug
xdebug.idekey = PHPSTORM
xdebug.trace_output_name=trace.%R.%u
xdebug.profiler_output_name=profile.%R.%u
xdebug.output_dir=/shared/xdebug
xdebug.log = /var/log/xdebug.log
xdebug.log_level = 3
; Try to discover the client host, otherwise fall back to the docker host
xdebug.discover_client_host=true
xdebug.client_host=host.docker.internal
; When you cannot specify a trigger, use "xdebug.start_with_request = yes" to autostart debugging for all requests
; https://xdebug.org/docs/all_settings#start_with_request
xdebug.start_with_request = trigger
; Set xdebug.mode trace to use this
; More details at https://derickrethans.nl/flamboyant-flamegraphs.html
xdebug.trace_format=3
xdebug.trace_output_name=xdebug.%R.%u
Apply changes by restarting apache in the container: apache2ctl -k graceful
Notes:
host.docker.internal resolves on Linux Docker only if the container was started with --add-host=host.docker.internal:host-gateway (nextcloud-docker-dev already does this).discover_client_host = true makes xdebug follow X-Forwarded-For - useful behind Nextcloud's dev reverse proxy.XDEBUG_SESSION=PHPSTORM php occ status inside the container and check /var/log/xdebug.log.
XDEBUG_SESSION cookie when toggled to Debug.
Click the Xdebug Helper icon in the browser and set it to Debug.
gbp buildpackge has succeeded finally on local machine,
but need to tidy and cleanup stuffs.
I didn't know packaging with Bazel best practice yet, to remove many third party vendor/ bundles, I've found that it requires pile of patch to eliminate them.
In the current version of Debian, as a one of build system, further work such as support from debhelper -
will be needed.
I'll file working progress on #1085173
Welcome to the June 2026 report from the Reproducible Builds project!
In these reports, we outline the most important things that we have been up to over the past month. As a quick recap about what problem our project intends to solve, whilst anyone may inspect the source code of free software for malicious flaws, almost all software is distributed to end users as pre-compiled binaries. The motivation behind the reproducible builds effort is to ensure no flaws have been introduced during this compilation process by promising identical results are always generated from a given source, thus allowing multiple third-parties to come to a consensus on whether a build was compromised or not.
If you are interested in contributing to the project, please visit our Contribute page on our website.
In this month s report, we cover:
A very interesting demonstration is now available showing how you might configure your Debian system to only install packages that have been reproduced by m/n rebuilders.
This is implemented via a reproduced+https:// APT transport ( a mechanism for communicating between the APT client and its repository source commonly HTTP):
Every package download is intercepted by repro-threshold, which queries two independent rebuilders for a signed attestation before allowing installation to proceed. [It] is important to note that [an] install will only succeed if all package dependencies are also reproducible.
The demo gives examples of how to quickly experiment with this using a Docker container.
In Debian this month:
debian-installer package in Debian was uploaded with a substantial reproducibility-related changelog. This means, for the first time, the uploaded version could finally be reproduced.
JDK-8385738 ( Javadoc does not produce reproducible output ) (for example). [ ][ ][ ]
reproducible-notes.git repo.
The IzzyOnDroid Android APK repository reached its next milestone this month, now covering 2 out of every 3 apps (66.7%) with reproducible builds. Their documentation for debugging and fixing failed builds has steadily grown as well. More clients have picked up showing reproducibility results (e.g. Droid-ify), and Neo Store now can be configured to stick to only reproducible applications. Further, an independent builder has been added to the build farm, increasing the trust level even more as APK builds can have multiple confirmations now.
At the same time, IzzyOnDroid s rbtlog got several new features. The most outstanding is caching for frequently used resources such as reproducible-apk-tools, command-line tools and NodeJS in order to counter ongoing issues with GitHub availability, while at the same time saving bandwidth and build time. This change also enables some other some smaller enhancements such as being able to configure build timeouts per recipe for those builds running longer than the average, release pattern filtering for update checks or having a field for maintainer notes to shortly summing up e.g. why a reproducible build failed.
Lastly, Bernhard M. Wiedemann posted another openSUSE monthly update for their reproducibility work there.
diffoscope is our in-depth and content-aware diff utility that can locate and diagnose reproducibility issues. This month, Chris Lamb made the following changes, including preparing and uploading versions 319, 320, 321, 322 and 323 to Debian:
Flags: line in the output of ocamlobjinfo, so adjust the test for cross-distribution compatibility. [ ]--long-form-style arguments when calling apktool in order to support apktool version 3. [ ]os.path.commonprefix method [ ].
In addition, Vagrant Cascadian updated diffoscope in GNU Guix to version 321 and 323.
/usr/bin/strip-nondeterminism. (#1139000)debian/watch format. [ ]Rules-Requires-Root: no and Priority: optional fields. [ ]Standards-Version to version 4.7.4. [ ]npm package includes an (automatically executed) preinstall script that is an ELF binary.
The change updates dependencies to the latest versions, and adds support for multiple -H options, to query results from multiple rebuilderd instances. The results are also now fetched concurrently.
debian:sid container now is 100% reproducible.
itertools and zip Python dependencies.
Yet again, there were a number of improvements made to our website this month including:
Date methods. [ ]
--release argument for consistency. [ ]
node-fuse.js.node-egjs-hammerjs.node-chartjs-adapter-date-fns.mkdocs-include-markdown-plugin.lmarbles.rocm-docs-core.libecoli.golang-github-tobischo-gokeepasslib.c-munipack.latex-coffee-stains.cxxtest.slime.spooles.sdpb.procmail.proftpd-dfsg.mah-jong.dxf2gcode.afterstep.ledger2beancount.ocamlviz.
Kenichiro Muto and Kuniyasu Suzaki of the Institute of Information Security in Yokohama, Japan published an interesting paper this month titled Attestable Build Chain: Enabling Trust in Reproducible Builds (PDF). Their abstract is as follows:
Ensuring trust in software supply chains requires verifying not only artifacts but also the processes that produce them. Although Reproducible Builds (R-B) require rebuilding to validate artifacts, they cannot verify whether the build was executed with the intended toolchain and inputs and may reproduce unintended or compromised builds without detection. We present Attestable Build Chain, a framework for externally verifying build-time execution without rebuilding. Rather than preventing compromise, it provides verifiable, tamper-evident evidence of actual build-time execution, enabling verification of build process integrity from observed file accesses during the build. [ ]
Julien Malka, Stefano Zacchiroli and Th o Zimmermann published a 50-page report detailing A Decade of Software Reproducibility in the Nix Package Ecosystem:
We find that functional package management enables extremely high rebuildability over time (near-universal ability to reconstitute historical build environments and rebuild software packages), while bitwise reproducibility has steadily improved and reaches a high point in recent years (up to 93% in 2024). Early years show substantially lower bitwise reproducibility, indicating that functional package management alone does not guarantee bitwise-identical outputs, and that the observed high level of bitwise reproducibility is not solely due to the package management approach. Common causes of unreproducibility, both in the rebuildability and bitwise reproducibility dimensions, include management of dates in build and test processes; we quantify their prevalence and other common causes using manual analysis of logs of rebuild failures and automated analysis of diffoscope.A PDF of their report is available online
Tim Bastin of L3montree GmbH and Jacek Galowicz of Applicative Systems GmbH from DevGuard published a paper detailing How We Built a Sovereign, Reproducible Container Supply Chain for DevGuard:
This paper presents how the DevGuard project rebuilt its OCI container pipeline around reproducible Nix builds and independent dual-platform digest verification. DevGuard images are built hermetically from pinned source revisions, signed with Sigstore/Cosign, and verified through digest comparison across GitHub Actions and sovereign GitLab infrastructure hosted on container.gov.de. We describe the practical integration of reproducible OCI image builds into existing CI/CD workflows and argue that independently reproducible container digests provide a stronger integrity guarantee against build tampering than provenance alone. The paper further discusses remaining trust assumptions and the relevance of sovereign build infrastructure for government and regulated environments.
Finally, Yiseul Choi, Junga Kim, Jun-Ho Hong and Seongmin Kim of the Department of Convergence Security Engineering at the Sungshin Women s University in Seoul, Korea titled Attestation-based verification of SBOM integrity via consumer-side reproducibility:
Software bills of materials (SBOMs) support supply chain transparency, but they do not prove that a delivered SBOM reproducibly corresponds to its software artifact. Existing signing and provenance mechanisms protect integrity and traceability, yet lack consumer-side reproducible verification. We propose an SBOM integrity verification framework combining procedure disclosure, consumer-side reproduction, authority-generated reference evidence, and digest comparison. A trusted authority records a reference digest, and consumers compare it with locally reproduced and delivered SBOM digests. Experiments on 100 real-world container images show detection of artifact tampering, SBOM substitution, distribution modification, and adaptive tampering beyond signature-based approaches
#reproducible-builds on irc.oftc.net.
rb-general@lists.reproducible-builds.org
a.mx.mayfirst.org, b.mx.mayfirst.org and c.mx.mayfirst.org) instead of
our name servers (a.ns.mayfirst.org, b.ns.mayfirst.org and
c.ns.mayfirst.org). The weird part: this mistaken configuration was happening
at the registrar level, protected by each member s own credentials that we
don t have access to.
Each affected member fixed their records to resolve the problem but also made
very clear that they had not logged into their registrar in years, sugggesting
that the DNS authoritative records in their registrar accounts spontaneously
changed on their own. The first time was weird, the second time could possibly
be a coincidence? But by the third time this happened, we started to panic.
How could registrar records spontaneously change? All three domain names were
registered with different companies - so it couldn t be a single registrar
problem? Are we going to get a flood of these complaints? What is going on!?!?
We did an inventory to see if this was happening with other domain names in use
by our membership and that s when we discovered just how hard it is for our
mostly non-technical users to set a domain s authoritative name servers. The
error rate was less than 1% but still that was a lot of domain names with
typos:
a.ns.mayfist.org!a.ns.matfirst.orga.ns.mayfirst.org, b.ns.mayfirst.org, c.ns.mayfirst.org,
a.mx.mayfirst.org, b.mx.mayfirst.org, c.mx.mayfirst.org, and even
a.webproxy.mayfirst.org - in other words, all the domain names we tell you
do to anything with.a.mx.mayfirst.org, b.mx.mayfirst.org and c.mx.mayfirst.org.c.ns.mayfirst.org and c.mx.mayfirst.org. It hangs out
in case something terrible happens to our main colo. The week before we started
receiving these complaints, I separated these services, moving
c.mx.mayfirst.org to a dedicated MX server. As a result, these two domain
names stopped pointing to the same IP address. And that s when the complaints
started rolling in. In other words: the affected members set the incorrect name
servers years ago, but because just one of the name servers resolved to an IP
that happened to provide the correct authoritative lookup services, it went
undeteced all this time.
So mystery solved. Nobody s authoritative registrar records suddenly
changed. They were mis-configured for years but thanks to the amazing
resilience of the DNS system, nobody noticed because just one working DNS
server is all you need.
I used to ice skate as a teenager but I stopped at University. I tried to
pick it back up in 2024 but had to stop when I got ill. I restarted
in 2025, initially with a weekly skate session but last month I started group
hockey skate lessons.
I've been skating in a pair of Bauer1 Nexus N77s that I bought 7 years
ago on a work trip to Toronto.
These did a great job of getting me back into the hobby for 6 years but
recently I felt it was time to step up to a better quality pair.
Despite being a size down from my shoe size, the Nexuses are too
large: I had been compensating with thick socks but still struggling to
get the boots tight enough. I'd have to wear gloves to lace up because I'd
cut my hands pulling the laces otherwise.
After too long researching/deliberating/kvetching (very much on trend for me)
I upgraded to Bauer Vapor Fly30s another half-size down (and nearly ten times
as much).
The fit is much better, in almost every respect. They actually go on easier
and I don't have to tear my hands tightening the laces. They feel like a natural
extension of my feet. I seem to be using a different set of muscles to skate,
so the first few sessions were very fatiguing, but that settled.
The Vapor line is speed-oriented, which I thought would fit my skate style best.
I have unfortunately gained a common problem: arch pain. More precisely, my
navicular bone seems to be quite prominent2, and that part is
pressing uncomfortably into the boot. Boots typically take a few sessions to
break in, but after 7-8 sessions the pain was getting to the stage that I
couldn't skate for a full session without being in agony.
The last time I skated I tried to throw everything at the problem: I'd had
the skates baked3; bought some orthotic insoles; then some "Bunga" pads over the sore bit and an
attempt to more loosely tie the laces over the affected area. I tried a ten
minute skate, and it seemed a bit better.
I then tried experimentally to swap back to my old skates, and I felt like
Bambi: I just couldn't do it!
They didn't press on the navicular, and they're
softer so you can compensate for the size with tight lacing, but I had no
confidence in them, I couldn't lean into the turns. They just felt weird.
I realised there's no way back.
I switched back to the fly30s, adjusted the bunga pad positioning, tweaked
the lacing and went back on for about 40 minutes. It went well: the rink was
quiet, it was cool whilst we had a heat wave outside, so I worked up a sweat.
By the end there was some discomfort, but not too much, and I think partly
the area is currently sensitive so just about anything will cause discomfort.
Fingers (or toes) crossed that I've mitigated the problem! If not, it might
be time to try a punch out.
bbc.co.uk##aside
bbc.co.uk##footer>div:has(h2)
bbc.co.uk##[data-block="uploaderEmbed"]
bbc.co.uk##[data-block="links"]
Earlier this year, I made myself a new hat (it will be blogged), and I
wanted to put a nice cockade on it.
So, I looked for suitable ribbons, and couldn t find any.
I found some ribbon that looked passable, but the colours I wanted
weren t available, so I set up the website to notify me, and kept
working on the hat.
Eventually the hat was done, the ribbon was still not available, so I
decorated the hat with fake flowers from my stash, and started wearing
it.
And that s when I got notified that the ribbon was back in stock.
By the time the ribbon arrived, I had decided that the hat looked better
with the flowers, and project cockade was put on hold, possibly for a
future hat.
And then June came, and what could be a better time than that for a
project based on flag colours?
As I feared, the ribbon I got wasn t the best: it was a bit too stiff
and plasticy, and not really usable for many other things. It did
however work well enough for a cockade, and I decided it was a good
chance to try different methods and designs, and then make another one
to take pictures and publish step-by-step instructions.
And that s the perfect recipe to find oneself surrounded by a somewhat
unreasonable number of very similar cockades, I guess.
I looked around for instructions, and the ones that gave a result that
was closer to my mental idea of a cockade were the ones by American
Duchess,
so on my first attempt I tried to follow those.
I didn t have a cork board with a hole, so it was a bit fiddly, and the
result was passable, but could have been better. Meanwhile I saw a
forum post commenting on the above tutorial
and that gave me ideas for a procedure more suitable to the tools I had.
The second cockade I made was indeed more satisfactory, and I also
started to experiment with making a center piece with ribbons, to cut
down on the types of materials needed.
On this one I also tried to add a pin backing, so that I could write
instructions on how to do it in what I believe is a more stable way than
simply adding it to the felt backing at the end.
I m not sure whether the other ones will be tacked to a hat, and thus
won t require pins at all, or if I ll just put them on some dress with
pins hidden under the ribbon layers.
And then I was ready to make a third cockade, taking step by step
pictures for my website, and I planned to start on it the next morning.
Trenord had different ideas.
Thanks to the combination of independent but complete disruptions on
two nearby train lines, I spent the morning driving a couple of people
to the nearest station that was still being served by trains, and then
back home less than 10 minutes before I had to start working, which if
you ask1 me was pretty homophobic of the train company.
There was way less traffic than I expected, and I did enjoy the drive
2, but for various reasons it meant a significant delay for this
post.
Anyway, less than a week later than I had planned, halfway in June I
managed to publish step by step instructions on my website, but I wasn t done with the project yet.
Beside the fact that I still needed to finish sewing the backing felt to
the cockades I had done, I also had a few ideas for more centrepieces
made of ribbon I wanted to try.
And this means that I moved on to another project that was already in
progress (this one will also be blogged).
After I ve finished that one, at the very end of June I quickly made the
last two centrepieces, taking pictures for the instructions, and in the
next few days I also finished the cockades.
This time, instead of plain pleats I tried to use box pleats, and I
quite like the look they give, so if in the future I ll have a need for
more cockades I may use again this pleating pattern.
For the last cockade I wanted to try two things: putting a pin in the
middle as a centrepiece, and gathering the ribbons.
For the pin, I found that the only one I had that had a colour scheme
compatible with the ribbons I had was one with the penguin from
linux.it, which had a black background, so I put black ribbon on the
outside and white next to it for contrast.
And gathering was done with a whipped gather with ribbons that were one
and a half times the outer circumference of their slot, and looks decent
enough, but I think I prefer the look of pleated cockades a lot. Maybe
it would look better with a softer ribbon.
Anyway, I think this is plenty of cockades for the time being, unless I
get tempted by buying more colours of ribbon to make different ones. But
I m not making an online purchase just for those. I am not.
Taking a good group photo consists of multiple aspects:
| Series: | Culture #2 |
| Publisher: | HarperPrism |
| Copyright: | 1989 |
| Printing: | February 1987 |
| ISBN: | 0-06-105356-2 |
| Format: | Trade paperback |
| Pages: | 295 |
Team Rcpp is excited to share that an brandnew new version 1.1.2 of
Rcpp is now on CRAN, has also
been uploaded to Debian, and has
already built for r2u
and r-universe;
Windows etc builds at CRAN
should follow in due course.
Rcpp has long established itself
as the most popular way of enhancing R with C or C++ code. Right now,
3236 packages on CRAN depend on
Rcpp for making analytical code go
faster and further. On CRAN, 13.4% of all packages depend (directly) on
Rcpp, and 61.4% of all compiled
packages do. From the cloud mirror of CRAN (which is but a subset of all
CRAN downloads), Rcpp has been
downloaded 121.6 million times. The two published papers (also included
in the package as preprint vignettes) have, respectively, 2263 (JSS, 2011) and 471 (TAS, 2018)
citations, while the the book (Springer useR!,
2013) has another 742.
The is the second update in the 1.1.* series which had, among other
changes, switched to C++11 as
the minimum standard. This release continues as usual with the
six-months January-July cycle started with release
1.0.5 in July 2020. Interim snapshots are always available via the
r-universe page and
repo. We continue to strongly encourage the use of these development
released and their testing we tend to run our systems with them too.
Having said that, we would like to reiterate that we strongly object
to the upstream R release and change management which in this 4.6.*
cycle made several abrupt changes forcing packages which
consume header files to make very abrupt change. Rcpp, just like numerous other CRAN packages demonstrates that
API changes can be undertaken responsibly in a managed manner which
allows for transition periods followed by possible warning periods,
deprecations periods and finally (but only at long last) errors. What
happened here is a speed run to the final stage of forced errors. Uncool
and irritating for something as widely used as R. This forced us to make
an interim release 1.1.1-1.1 even though we have of course had a policy
of always keeping properly tested, installable, and error-free
releases candidate version in the main repository branch and hence
available via R-universe tested
packages for all relevant platforms, and even via binaries for most
(including Ubuntu LTS). It would be nice if R Core found a way to take
advantage of this. Maybe development cycles, running apart for a year as
they do for R, should also include selected packages.
Once again I am not attempting to summarize the different changes.
The full list follows below and details all these changes, their
respective PRs and, if applicable, issue tickets. Big thanks from all of
us to all contributors!
Thanks to my CRANberries, you can also look at a diff to the previous interim release along with pre-releases 1.1.1-1 and 1.1.1-1.1 that were needed because R-devel once again sudden decided to move fast and break things. Not our doing. And there also should not have been a need to two such uploads but it was amateur hour all around. Questions, comments etc should go to the GitHub discussion or issue section, or the Rcpp list. Bugs reports are welcome at the GitHub issue tracker as well. GitHub offers decent search for issue, pull requests and discussions; as many topics have been covered it is worth checking as well.Changes in Rcpp release version 1.1.2 (2026-07-01)
- Changes in Rcpp API:
- Use of
execinfo.his again conditional to avoid build complexity (Dirk in #1445 addressing #1442)- An internal state component for
Datetimeis nowint(Dirk in #1448 and #1449 fixing #1447)- Three new (in R 4.6.0) attribute accessors are used conditionally (Dirk in #1450 closing #1432)
- An UBSAN error in the Sugar-based NA comparison has been corrected (I aki in #1453 fixing #1452)
- Treatment of Inf outside of integer range in Sugar function has been corrected (I aki in #1458 fixing #1455)
- Integer overflow protection has been added for sugar functions (I aki in #1457 fixing #1454)
- The parent environment is now accessed via
R_ParentEnv(Dirk in #1460 fixing #1459)- Change to returning
dataptragain for better handling of empty vectors (I aki in #1462 fixing #1461)- Undefined behavior errors in use of
ListOfproxies have been addressed (I aki in #1464 fixing #1463)- Under newer R version,
R_UnboundValueis no longer used (I aki in #1466 fixing #1465)- New R API access point
R_getRegisteredNamespace()is used with current R versions (Dirk in #1469 fixing #1468)- The
Nullable::as()exporter now uses an explicit cast to the templated type (Dirk in #1471 fixing #1470)- A memory leak in the variadic
Rcpp::warning()template has been fixed (Kevin in #1475 fixing #1474)- The
Nullable::operatorT()has been added as a 'opt-out' (Dirk in #1477 with coordination in #1472)- Add templated integer-index overload for
operator[]on small systems such as WASM (Jeroen Ooms in #1482)- The attribute accessors in
AttributeProxyPolicyno longer rely onget__()(Kevin in #1484 fixing #1483)- Changes in Rcpp Documentation:
- Reference in the bibliography used by the package vignettes have been updated.
- Changes in Rcpp Deployment:
- Excute permissions are set consistently on scripts with shebangs (Mattias Ellert in #1467)
- R 4.5.* has been added to the CI matrix (Dirk in #1476)
- Three nag messages issued when obsolete build flag accessors are used now show
Rcpp:::(Dirk in #1480 fixing #1456)- Reference GitHub Actions have been updated to their current versions (Dirk in #1481)
- Non-release Changes:
- A non-release hotfix 1.1.1-1 used by CRAN accommodates breaking changes to the API in R 4.6.0. It would be nice to have the same level of release management in R itself that CRAN expects from us.
This post by Dirk Eddelbuettel originated on his Thinking inside the box blog. If you like this or other open-source work I do, you can sponsor me at GitHub.
As is the case for many people working in the security industry, the last
few months of my life have been focused on dealing with people wanting to
use LLMs everywhere. From an enterprise security perspective that s not an
inherent problem - what s more of a problem is that people want those agents
to have access to resources like their calendar and email and so on, and now
we have somewhat non-deterministic agents that seem very enthusiastic to
achieve what you asked whether that s a good idea or not, and we re
combining this with credentials that give them access to sensitive data, and
leaving those credentials on disk where they can be committed into git repos
or exfiltrated to some other service to make use of them on the agent s
behalf or well just any other number of things, at which point your CEO s
email is suddenly readable by everyone and you re having a bad day.
As I mentioned in my last
post, pretty
much every strong mechanism for keeping credentials in place is just not
supported in the wider world. We can imagine a universe where agents use
hardware (or at least hypervisor) backed certificates to obtain credentials
and any that end up leaking are worthless as a result. But, sadly, that s
not an option for most people using existing identity providers. The state
of the art is that you use the device code
flow and a human authenticates and
the token ends up back inside the agent environment and then it proceeds to
do whatever it wants with it and you just hope that you wake up the next
morning without an awful infoleak occurring.
(An aside: I do not like the device code flow as used in enterprise
environments, and I never will. The identity provider doesn t have a real
opportuity to inspect the security posture of the system asking for the
token, and as a result some identity providers will restrict tokens that are
issued in this way. The common alternative of doing stuff using a more
standard flow and having a redirect URI pointing at localhost works fine for
local systems and is a pain for remote ones, even if you can commit crimes
with SSH forwarding. I m going to suggest something that I think is better,
and you are free to disagree)
I m not in a position to get every identity provider and service provider to
change their security posture, so I m somewhat stuck in terms of the tokens
they re willing to issue me - largely either JWTs or opaque access tokens,
with no support for any mechanism of binding that token to an instance. The
token that s going to have to be provided to the remote service is something
I have little influence over. But that doesn t mean I can t influence the
token that lands inside the agent s environment. I can issue a placeholder
token to the agent, and force it to communicate via a proxy that swaps out
the placeholder for the real thing. The worst the agent can do is exfiltrate
the placeholder token, and as long as malicious actors don t have access to
that proxy, it doesn t matter - nobody else can do anything with the
placeholder.
This isn t a terribly novel insight, and it seems like almost everybody has
reinvented this on their own. But a lot of these implementations involve you
somehow obtaining the real token in advance and then pasting that into
something that generates a placeholder that you provide to your agent
environment somehow, and it s all a bit clunky and awkward, and it also
means that you need to deal with something that keeps track of the mapping
between placeholders and real tokens and oh no we ve just invented a secret
store, and if you want this to work at scale and reliably you re just
invented a high availability distributed secret store, and a lot of people
who ve read that are now shaking their heads and reaching for gin. Can we
simplify this, and improve security at the same time? I think we can!
Remember when I said as long as malicious actors don t have access to that
proxy, it doesn t matter ? What if they do? What if they compromise one
machine inside your environment and are then able to email a bunch of
employees and convince their agents to send more tokens back to them and
then delete the email before a human reads it? Now you have someone inside
the wall with access to those tokens, and presumably with access to the
proxy, and now they can be anyone whose agent was gullible enough to think
sending them a token was a good idea. This isn t good!
So, I thought for a while, and I came up with a new idea. We can have a
broker service that obtains credentials for us. We can run that centrally,
away from the agents. A client in an agentic environment can request a
token, and that can result in a URL being generated and the user being
directed to open a URL in a browser and authenticate. When the user
authenticates, the authentication flow redirects the confirmation back via
the broker, and the broker obtains the real auth token. The obvious thing to
do now would be to return the auth token to the client in the agentic
environment, but we don t do that. Instead, we mint a new JWT, and add a new
claim - one that contains an encrypted copy of the token. In the process we
can copy over all the original claims, because those aren t secret - and now
even if the client inspects the token to figure out what access it has,
it ll get a correct answer. We sign the new token with our own signing key,
and pass that back to the client. The client now has a legitimate JWT that
is utterly useless, because the signature isn t trusted by anyone other than
us.
How does it use it? It makes an API request via a proxy, including the new
token in the Authorization: header. The proxy verifies the signature on the
token, and then decrypts the original token and swaps out the fake token for
the real one. The remote API sees what it expects, and everyone is
happy. There s never a real token in the agentic environment, but also we
don t need to store anyting anywhere. The only state is the encryption keys,
and those can be injected into the environment at startup. You need to
scale? Just start more of these processes. You need to support multiple
availability zones? Just start more of these processes in different
places. No persistent data is ever held in the broker or the proxy. You
don t need to care about distributed databases or secret stores.
This felt wonderfully elegant and I felt smug about coming up with a better
idea, and then I went to a bar earlier this week and sat down to read RFC
8705 and the guy next to me
saw that over my shoulder and asked what I was reading and I explained why I
was interested and we talked about agentic identity and then he mentioned
that fly.io had something that sounded very
similar and I read that and gosh yes
it is very similar, so damn you fly.io for stealing my ideas 3 years before
I even had them. Anyway. Now I need to do better.
Remember that there s still a risk around anyone who has access to the proxy
having access to the encrypted keys? We can remove that risk as well. It s
not uncommon for agentic environments to have an identity issued via
something like SPIFFE, at which point they have a
client certificate. You can probably guess where I m going with this. If we
require that an agent present a client cert to the broker when requesting a
token, we can embed a representation of that client cert into the token we
mint. The proxy can then require mTLS for the client connection, and can
verify that the presented certificate matches the one represented in the
token. If it does then whoever s using the token has access to the private
key associated with the environment it was issued to. If we then ensure that
the private keys backing these certificates are either hardware or
hypervisor backed, and as such tied to a specific instance, we now have a
high degree of confidence that the token can only be used in its intended
environment. Even if our identity provider doesn t support RFC 8705, we can.
This is fairly straightforward where you re using a platform where your
identity provider is also the environment that s consuming your tokens, and
more annoying for third parties. The broker potentially needs some amount of
third party vendor knowledge to make that work for everyone. This is even
more the case where login isn t via your identity provider (thanks, github),
but none of this is insurmountable - just annoying. And where vendors issue
opaque tokens rather than JWTs, this still isn t a problem; we can just mint
a new JWT that includes the opaque token as an encrypted claim, and include
the same certificate binding. The opaque token ends up being the thing
that s presented to the third party, but only after we ve verified the mTLS
binding.
In an ideal world none of this would be necessary - someone would spin up a
new agentic environment, a user would prove their identity, and a
certificate embodying that identity would be issued to the environment with
a private key that can t be exfiltrated. That certificate would be
sufficient to obtain new certificates associated with the same private key,
and we could still bind that into mTLS identity. This would be much simpler,
but browsers don t support it, so it s not likely to happen any time soon.
Anyway. Even if we can t have the best thing, we can do better than we are
at the moment, and also it would be lovely if we could standardise on this
rather than have everyone build their own thing. The end.
Next.