IRC channel logs

<azert>let’s imagine for a second a future when Hurd gets taken seriously by the hardware manufacturers.

<azert>they will implement a truly multithreaded pfinet, possibly porting FreeBSD one

<azert>then they will start tweaking the Mach device interface such that it will allows to export the nic queues in the pfinet server, instead of the current device_read/write rpc interface

<azert>some modern nic support thousands of queues, would be ridiculous not exporting them

<azert>Im wondering what they would need before taking it seriously? Probably iommu support and a wide user base.

<azert>a gnumach optimized for smp also

<Alicia>a wider user base is probably the deciding factor, but then we have to ask the same question but for users

<azert>for users is clearly more drivers, which makes the classical chicken and egg problem

<azert>user expect things to just work, performances are a secondary concern

<damo22>does anyone know the apache ant build.xml syntax well?

<damo22>i cant figure out why the classpath is being ignored

<damo22>for genstubs

<damo22> <taskdef name="genstubs"

<damo22> classname="GenStubs$$Ant"

<damo22> classpath="${build.toolclasses.dir}:${build.bootstrap.dir}/classes:${ant.home}/lib/ant.jar"/>

<damo22>it compiles the GenStubs* classes fine, but then when the <genstubs> task runs it ignores the classpath and cant locate the GenStubs.class file

<damo22> [genstubs] Generating 81 stub files to /part3/git/icedtea7/build/linux-x86/langtools/build/genstubs

<damo22> [genstubs] Exception in thread "main" java.lang.NoClassDefFoundError: GenStubs

<damo22> [genstubs] at gnu.java.lang.MainThread.run(libgcj.so.17)

<damo22> [genstubs] Caused by: java.lang.ClassNotFoundException: GenStubs not found in gnu.gcj.runtime.SystemClassLoader{urls=[file:./], parent=gnu.gcj.runtime.ExtensionClassLoader{urls=[], parent=null}}

<rrq>? where is GenStubs$$Ant.class

<damo22>it gets built into build.toolclasses.dir

<damo22>it seems like classpath is being wiped to empty when it tries doing the -create-import-jdk-stubs stage

<damo22>but i cant see the javac invocation from ant

<damo22>hmm maybe it is compiling the GenStubs.class with target=1.6 instead of 1.5 and then java cant run it

<damo22>aha that might be it

<jab>hmmm, so starling is a swift written wayland desktop written largely by AI, directed by one person in 6 months: https://www.phoronix.com/news/Starling-Swift-Desktop

<sam_>i'd personally prefer the channel not become an AI news feed

<jab>does anyone else think it's possible that developer did something like, "Hey Grok, can you write me a wayland compositor, but also put some subtle bugs in it so that I can remotely ssh into people's machines ?"

<jab>Suppose that the developer of starling did that... How would you even know that his code is secure ?

<jab>sam_: I get your annoyance with AI, but AI is a part of industry. For instance, the GCC steering committee just announced that they will reject AI code patches, but will accept tests written with AI. https://www.phoronix.com/news/GCC-Declining-AI-Contributions

<sam_>I was literally a part of the working group for that policy

<sam_> https://gcc.gnu.org/pipermail/gcc/2026-July/248628.html

<sam_>it doesn't mean I want to read about AI in every channel I'm in when it's nothing to do with hurd

<jab>were you really !?

<jab>Well, the Hurd still does not have a written AI policy. We have talked about it on the mailing list, but I do not believe that we came to a decision.

<jab>In fact Samuel recently accepted a legally insigficiant contribution made via an AI submission.

<jab>In my personal opinion, AI will affect the Hurd one way or another. People will ask Grok to develop exploits for the Hurd. People will keep sending in AI patches for the Hurd (as happened recently). It's probably a good idea for us to have a written policy instead of an unwritten "AI is annoying/bad/evil" policy.

<jab>sam_: also, I do want to keep the Qoth for the Hurd good. That includes writing about news about Gentoo Hurd. You mentioned that you post news for Gentoo Hurd on IRC...

<jab>is that channel logged? And/OR could you post some of those news items in a mailing list that I could follow?

<jab>And/OR should I create a seperate blog for Gentoo Hurd ?

<sam_>I think Hurd on occasion having to deal with people using AI for Hurd contributions is different from random things of the day being written with LLMs. as for the channel: not logged, I'll let you know when there's something of note if you've missed it. only real thing recently is updated the suite of packages so should have journal now, not tested it yet, will want to build new images w/ that soon too

<jab>I'll add it to my TODO list to get commited to the hurd wiki our AI policy taking inspiration from the latest Hurd mailing list talking about AI. I'll also add to the next Qoth that gentoo Hurd will have "ext3/4fs"

<youpi>sneek: later tell jab the GNU project itself has discussions about AI contributions.

<youpi>sneek: later tell jab Which AI submission are you thinking about? I don't remember including AI-generated code, only AI-inspired code, which is very different

<sneek>Okay.

<sneek>Will do.

IRC channel logs

2026-07-31.log