<daviid>ot -any one knows a free s/w to extract a pdf barcode <antipode>You can use a combination of 'procedure-documentation', 'module-for-each' and 'variable-ref'. There's also some web server demo for listing all modules but I cannot find it anymore. <dsmith>nckhexen: Ok, can you unban the bot please <dsmith>I do get this: :silver.libera.chat 482 sneek #whereiseveryone :You're not a channel operator <nckhexen>I'd set the ban for 8h since I was about to hit some hay :) <nckhexen>nckhexen: I wonder what it's trying to do. <nckhexen>It's been free to do so since localtime:32:17 <nckhexen>dsmith: I really need sleep. I'll leave you to it. If you need to take #guix off the join list temporarily to get it to join #guile, that's fine. Night! <dsmith>Ok. If it goes nuts I'll just stop it <nckhexen>dsmith: <So it's behaving now?> No change in behaviour, still behaves as soon as it's banned & unbanned, not before. <nckhexen>So it needs to be configured to (quick fix) stagger joins, one channel at a time, or (better) properly rate limit all messages, which is clearly not happening, no matter what the bot says. <Noisytoot>dsmith: What is it doing that causes it to flood off? <dsmith>Not sure yet. Seems like systemd keeps restarting it. <dsmith>Filesystem is full currently. Trying to clean that up. <Noisytoot>Excess Flood means it's exceeding its RecvQ or sending too many messages too quickly <dsmith>Well, near as I can figure, bobot is quitting becuse whois list is flooding the bot <dsmith>Only 292K bytes free! It's a 1.7G rootfs <dsmith>Well two things. The "seen" table is way too huge. Should only have one row for each nick. So some SQL code is buggy. <dsmith>Also, it seems bobot is sending many "MODE +b". Looks like for IPv6 <sneek>Not as far as I can remember. <sneek>dsmith was last seen in #guile 0 seconds ago, saying: sneek: seen dsmith. <sneek>dsmith was in #guile 0 seconds ago, saying: sneek: seen dsmith. <dsmith>Ok, fixed the bug in the "seen" db. But also removed all the old recods. A fresh start! <dsmith>No, that's me trying something.. <dsmith>For some reason, it's sending lots of "MODE #guix +b *!*@2001:470:69fc:105::1:6a20" like messages. <dsmith>Cool. And one more without debug info... <sneek>spk121 was last seen in #guile 12 minutes and 54 seconds ago, saying: cool, carry on. <dsmith>sneek: later tell nckhexen Just might have worked around it. <daviid>i am running GNU Guile 3.0.8.71-dadee-dirty, and i thought it included a patch that solves these errors: the Exception thrown while printing backtrace: In procedure "primitive-call-ip": Wrong type argument in position 1 (expecting "PRIMITIVE_P"): #<procedure 7ff1a4c22da0 (_ _ _ _ _ _ _)> <spk121>daviid: I don't see any likely-sounding commit in the git log, unless you go back to 2019 <nckhexen>dsmith: Thanks! The #guix banlist is hardly huge (22 entries, unless the bot is doing some weird/buggy subnetting thing to make those IPv6 entries explode), but it's bigger than it's ever been because a different bot now bans cryptocoin scammers on sight. That's what all those Matrix (2001:470) bans by libera/bot/litharge are from <nckhexen>Well, not subnetting, but like trying to match it against different users and requesting it each time? That doesn't make sense, but then I might be trying too hard to find a rational explanation for what's just a bug. I've not seen sneek track bans, so why does it need the list? <nckhexen>dsmith: I've removed most of those bans in the hope that they were permabanned on the Matrix side anyway, but this will happen again as time goes on and cryptos keep scammin'. <dsmith>nckhexen: The bot sneek doesn't need the bans, but the core bobot++ code did. <dsmith>nckhexen: Not needed with the *serv things on libera <nckhexen>Thanks for pointing that out, I was looking in the wrong place. How did you work around it? <dsmith>So for now, I've just commented out the code within bobot that responds to a "367". <dsmith>THe bot is back on #guix and seems happy <dsmith>The geeter is turned off. Had to wipe out the "seen" db. <dsmith>And I havn't re-enabled that other chan yet. <nckhexen>I read that. Sounds like you didn't find the bug? <sneek>I think I remember nckhexen in #guile 37 seconds ago, saying: #wie? Was it problematic?. <sneek>nckhexen was in #guile 0 seconds ago, saying: sneek: seen nckhexen . <dsmith>Yeh, it logs the command before it responds to it. ***dsmith is now known as not-dsmith
***not-dsmith is now known as dsmith
<dsmith>Ok greeter is back, and the other chan is back. ***dsmith-work is now known as dsmith
***dsmith is now known as dsmith-work
<old>sneek: later tell lampilelo that I added support for asynchronous mutex/condition-variable/barrier/semaphore and futures in guile-parallel. Almost x2 acceleration against ice-9 futures <lampilelo>old: that's great! i'm wondering though why did you put a dispatch in your semaphores, why not just expose wait-semaphore and post-semaphore? <sneek>Welcome back lampilelo, you have 1 message! <sneek>lampilelo, old says: that I added support for asynchronous mutex/condition-variable/barrier/semaphore and futures in guile-parallel. Almost x2 acceleration against ice-9 futures <old>lampilelo: I do this for semaphore and barrier to avoid having to check for type! <old>Returning a self-contained lambda is probably the beautiful abstraction I can think of. There's no need to check anything! Except for the dispatching itself for the semaphore <old>btw any though on what I could add to the library? <old>s/the beautiful/the most beautiful <lampilelo>but dispatching over 8 symbols will take longer than just checking for type <old>lampilelo: True. I will try with a post/wait-semaphore procedure instead to see if it improves the benchmark <dsmith-work>Silly bot, why don't you listen to bang commands any more? <lampilelo>old: and for suggestions: you could maybe add the concept of priority for jobs <lechner>Hi, is there a way to disable auto-compilation messages when calling from C via scm_with_guile? <lechner>Maybe I can just add --quiet to the existing command-line options via some invocation of scm_set_program_arguments_scm ? <sneek>daviid?, pretty sure was seen in #guile 15 hours ago, saying: spk121: ok thanks. <sneek>Its been said that Sneek is sneek: Sneek is sneek: chain reaction <sneek>Not as far as I can remember. <old>sneek: what is the meaning of life? <antipode>Sneek only supports a single word after 'what is' <sneek>I've heard meaning-of-life is 42 <sneek>Last time I checked some other thing is too many words <old>sneek: sneek is myself <old>sneek: what is sneek <sneek>I've heard sneek is many things. <old>sneek: what is sneek <sneek>From what I understand, Sneek is forgetful <old>did I get something wrong here? <antipode>old: I think it's just that sneek is many things, so it can return one of the many old entries for 'sneek' instead of the one you were intending. <old>Right. Would be cool to have an official documentation ^^ <old>sneek: where is your-documentation <old>sneek: later tell lampilelo that overhead improvement of semaphore on future seem marginal without the dispatching. But that's only for a simple benchmark <old>dsmith-work: Right. Sneek already told be about it ;-) <sneek>antipode was last seen in #guix 2 minutes and 52 seconds ago, saying: For example, coreutils has 'ls', 'cat', .... <sneek>old?, pretty sure was seen in #guile 3 minutes and 7 seconds ago, saying: s/be/me. <old>can't wait for sneek to take over cortana <sneek>Not as far as I can remember. <sneek>I last saw two[m] in #guile 2 minutes and 33 seconds ago, saying: :blob_thinking:. <old>lampilelo: fibers is around 10% faster right now from a quick benchmark