#archlinux32 | Logs for 2020-04-30

Back
[04:39:55] -!- DCyrax has quit [Ping timeout: 260 seconds]
[05:11:25] -!- torv has quit [Remote host closed the connection]
[05:12:33] -!- torv has joined #archlinux32
[07:01:35] -!- abaumann has joined #archlinux32
[07:01:35] <buildmaster> Hi abaumann!
[07:01:35] <buildmaster> !rq abaumann
[07:01:36] <phrik> buildmaster: <abaumann> I always knew, pentium is the slowest architecture ;-)
[07:40:50] <bill-auger> did you guys hit this sort of error with firefox v75? https://lists.parabola.nu
[07:40:51] <phrik> Title: [Dev] [PATCH] Iceweasel update to 75.0-1 (at lists.parabola.nu)
[07:41:22] <bill-auger> seems to be one of those oxidation problems
[07:59:32] <abaumann> Hi. IIRC I built with rust 1.42
[07:59:40] <abaumann> maybe it's now rust 1.43 already..
[08:01:03] <abaumann> Never encountered something like that.
[08:01:54] <abaumann> Firefox 35 needs a minimal version of Rust, but if rust would be to old, you would get another message like "at least rust x.xx required"
[08:07:55] <abaumann> mmh. your patch is not changing much to firefox, so..
[08:08:02] <abaumann> ..it cannot cause too much trouble.
[08:09:20] <abaumann> let me try a rebuild and see if I hit the same problem..
[08:15:56] <abaumann> a small D-mess: lldb is broken because llvm-libs 10 cannot be pushed to stable, because ldc is broken because dmd is broken because it cannot find libLLVM-9.so :-)
[08:16:26] <abaumann> @deep42thought: I'll try to fix that one..
[08:16:33] <girls> db-update -f ?
[08:17:05] * girls finally power-cycle their raspis
[08:17:15] <abaumann> morning ;-)
[08:17:21] <girls> morning :-)
[08:17:32] <abaumann> yeah. but what else breaks then?
[08:18:30] <abaumann> btw: https://archive.archlinux.org
[08:18:31] <phrik> Title: Index of /packages/l/llvm-libs/ (at archive.archlinux.org)
[08:18:52] <abaumann> upstream seems to clean up their archive quite agressively
[08:18:55] -!- deep42thought has joined #archlinux32
[08:18:56] <buildmaster> Hi deep42thought!
[08:18:56] <buildmaster> !rq deep42thought
[08:18:56] <phrik> buildmaster: <deep42thought> maybe, they auto-create those nice penguins during boot with imagemagick?
[08:19:01] <abaumann> compare that to https://archive.archlinux32.org
[08:19:01] <phrik> Title: Index of /packages/l/llvm-libs/ (at archive.archlinux32.org)
[08:19:47] <deep42thought> oops, I forgot to shut down my router
[08:29:48] <deep42thought> I don't mind keeping a longer history
[08:31:17] <abaumann> Now that you have encrypting disks.. :-)
[08:31:37] * abaumann is reminded to do something about his 92% backup..
[08:33:41] <bill-auger> the chroot has rust 1.39 - that is the version in extra
[08:35:00] <bill-auger> 1.42 is in staging - i wil try that
[08:37:00] -!- torv has quit [Remote host closed the connection]
[08:37:19] -!- torv has joined #archlinux32
[08:40:04] <abaumann> ok. because I think it tries to use an old rust version and then use rustup to bootstrap a newer version..
[08:42:27] <bill-auger> that would fail on parabola - the build tool disables the newtork
[08:42:40] <abaumann> good thing to do :-)
[08:43:57] <bill-auger> that error looks like abi change though - treating "which rustc" as two filenames
[08:44:35] <abaumann> which is a binary (package) or a shell-builtin?
[08:46:02] <bill-auger> or someone forgot to wrap`which rustc` as a shell call
[08:46:34] <buildmaster> i486/mesa is broken (says nlopc46-i486bs1): https://archlinux32.org
[08:46:44] <bill-auger> but it seems to be compiling now with rust 1.42, so thats good enough for me
[09:30:17] <deep42thought> maybe, we should disable the network during build() and package(), too?
[09:33:05] <deep42thought> bill-auger: how do you achieve disabling the network?
[10:21:25] <buildmaster> i486/linux is broken (says nlopc46-i486bs0): https://archlinux32.org
[10:31:45] * bill-auger fears that question was beyond my pay scale
[10:33:32] <bill-auger> based on a 30 second exploration, i beleive that is accomplished with the `unshare` command on LOC 129 https://git.parabola.nu
[10:33:34] <phrik> Title: libremakepkg ÂŤ chroot-tools ÂŤ src - libretools.git - Tools for working on Parabola (at git.parabola.nu)
[10:35:54] <bill-auger> networking is disable during prepare(), build(), and package() - only the fetch and validate phase can access the network; but that is because that stage happens outside the chroot - then the sources are copied into the chroot
[10:46:54] <buildmaster> i686/dmd is broken (says eurobuild6-2): https://archlinux32.org
[10:48:17] <buildmaster> i486/dmd is broken (says eurobuild6-7-i486): https://archlinux32.org
[10:56:24] -!- z3ntu_ has quit [Quit: killed]
[10:58:15] <buildmaster> i686/ldc is broken (says eurobuild6-2): https://archlinux32.org
[10:59:01] <buildmaster> i486/ldc is broken (says eurobuild6-7-i486): https://archlinux32.org
[11:11:10] -!- z3ntu_ has joined #archlinux32
[11:12:10] -!- dopsi has quit [Quit: ZNC - https://znc.in]
[11:15:12] -!- dopsi has joined #archlinux32
[11:19:44] <buildmaster> i486/nodejs are broken (says nlopc46-i486bs0): https://archlinux32.org
[12:06:20] <buildmaster> i486/mesa is broken (says nlopc46-i486bs0): https://archlinux32.org
[13:19:35] -!- buildmaster has quit [Remote host closed the connection]
[13:21:04] -!- buildmaster has joined #archlinux32
[13:21:04] <buildmaster> !rq buildmaster
[13:21:05] <phrik> buildmaster: <buildmaster> I might be insane, but never confused ... ;-)
[13:21:08] -!- trotz has joined #archlinux32
[13:22:01] <buildmaster> i686/d-containers are broken (says nlopc46): https://archlinux32.org
[13:22:38] <abaumann> @deep42thought: wc -l forced-package-builds.nlopc43 = 19 ?
[13:22:47] <abaumann> :-)
[13:23:17] <abaumann> building emacs 19 times makes it run much better afterwards.. ;-)
[13:26:09] <trotz> 2020/04/30 13:25 OK jeti100 PING PING OK - Packet loss = 0%, RTA = 16.07 ms
[13:27:14] <buildmaster> pentium4/d-containers are broken (says nlopc46): https://archlinux32.org
[13:32:43] -!- yans has joined #archlinux32
[13:48:06] -!- yans has quit [Quit: chaos is the only true answer]
[13:48:13] <deep42thought> bill-auger: thanks for the hint, I'll take a look, if we can do something similar (and possibly even upstream this)
[13:48:28] <deep42thought> abaumann: hmm, I guess, it's time to start that build-slave again ;-)
[13:51:21] -!- yans has joined #archlinux32
[13:55:09] <buildmaster> i686/psensor is broken (says nlopc43): https://archlinux32.org
[14:05:19] <deep42thought> abaumann: that's my usual approach, when I try to hack something into the chroot during build: I schedule it a couple times to my slave and if I fail to do my hack, I have still n-1 trials left :-D
[14:13:07] <deep42thought> hmm, changing makechrootpkg to disallow network access during prepare(), build() and package(), but *not* during check() is quite hard
[14:48:00] <bill-auger> it would be excellent if arch would do that but im quite sure they wont - i have discussed this with eli before
[14:48:44] <bill-auger> the problem is that many packages will require the network at build time - python pip thingsm java things, golang things, and the like
[14:49:45] <bill-auger> ideally, all of those dependencies should not be installed from the wild like that - they should all be packaged properly; but for some programs ive seen, that could be like 100+ new packages, only to support one useful program
[14:52:33] <bill-auger> i.e. those programs are actually not worth the effort of maintaining properly 0 they are only worth the effort if they are maintained lazily, with a blind eye to random dependencies pulled from the current HEAD of 100 github repos, like their upstreams expect
[14:53:47] <bill-auger> or "whatever a call to pip reterns today" - which could be different from what you built yesterdaym or 10 minutes ago
[14:55:17] <bill-auger> thats how i remember eli explaining it - it would result in many packages being dropped, some of them failry essential to the system
[14:56:01] -!- yans has quit [Ping timeout: 264 seconds]
[14:58:08] <deep42thought> yeah, I thought along that lines, too
[14:58:22] <bill-auger> if check() fails without the network, we prefer to skip it - either way it not a ideal
[14:58:26] <deep42thought> I think, one should argue with the reproducibility of builds
[14:58:56] <deep42thought> if you need network during prepare(), build() or package(), your result *will* be dependent on some source which you do not have under control
[14:59:05] <bill-auger> yep that is probably the reason why arch is not fully reproducible - anthraxx tried
[14:59:25] <deep42thought> I find this sad
[15:00:12] <deep42thought> imo, a distribution should try to be reproducible - otherwise, providing the sources alongside the packages is a nice gimmick which is not thoroughly useful
[15:00:27] <deep42thought> well, that may be a little too harsh ...
[15:01:32] <bill-auger> well people who use parabola are strongly opinionated about that - they want everything built from verifyable sources
[15:02:21] <anthraxx> bill-auger: hm?
[15:02:22] <bill-auger> after-all, source code that you do not have to inspect before compiling, is the same as magic blobs
[15:02:56] <anthraxx> deep42thought: strictly speaking you also dont control github tarballs, a re-tag can change that.
[15:03:07] <deep42thought> yes, but you will notice
[15:03:12] <bill-auger> oh i did not realize you were in this channel - apparently i typed your nick correctly from memory
[15:03:20] <deep42thought> :-)
[15:03:50] <bill-auger> a re-tag would fail the checksum though
[15:03:59] <anthraxx> true, but if its pulling in non deterministic sources you will notice via unreprducibillity and cry at upstream :D
[15:04:19] <deep42thought> yes, but you notice *after* building
[15:04:36] <deep42thought> imagine, I want to build for another architecture - say 32-bit
[15:04:43] <bill-auger> no, makepkg will bail before prepare even runs
[15:04:49] <deep42thought> then I have no way to find out it failed due to changed sources
[15:04:59] <anthraxx> sure. you can also go the way and have a package for all go, rust, npm and java deps, but thats a massive time and resource investment :)
[15:05:03] <deep42thought> if the source is checksummed, yes
[15:05:14] <deep42thought> but if it pulls in stuff during prepare() or build(), it won't
[15:05:24] <anthraxx> bill-auger: yes joined when starting to work with deep42thought to get better out of the box devtools experience for downstrea
[15:05:35] <bill-auger> yes i mean very strictly to avoid using VCS and SKIP
[15:05:39] <deep42thought> yes, I see that argument
[15:05:55] <anthraxx> bill-auger: VCS is fine as long as you use object hashes
[15:06:11] <deep42thought> I think, the current way is the right way in a big picture, but still: having reproducible sources would be *soooo* nice
[15:06:22] <anthraxx> i agree :)
[15:06:51] <deep42thought> there should be an easy way to stuff "everything downloaded during running `pip install whatever`" into the sources=() and properly checksum this
[15:08:12] -!- DCyrax has joined #archlinux32
[15:08:22] <anthraxx> will still be a massive investment, but it would be sweet to just not have networking during build
[15:08:33] <anthraxx> we have some golang libs for similar purpose https://www.archlinux.org
[15:08:34] <phrik> Title: Arch Linux - Package Search (at www.archlinux.org)
[15:08:46] <anthraxx> debian also went down the patch and has packages for every single rust crate
[15:09:06] <deep42thought> they could be auto-generated and auto-maintained
[15:09:17] <deep42thought> I do this with a bunch of perl and python packages
[15:09:49] <deep42thought> but of course, "automaintained" is only relative: every so often, something breaks ...
[15:10:13] <bill-auger> yea i was working with perl recently, and discovered a program that pulls from the perl repos and creates an arch package automatically
[15:10:45] <bill-auger> something similar could be made for pip and rubygems, that could go a long way
[15:10:54] <bill-auger> java and golang are pretty hopeless though
[15:11:59] <deep42thought> for ruby, there is [quarry]
[15:12:48] <anthraxx> in not much fan of automaintained, it most of the time lacks proper optional things like completions, man pages, docs and what not
[15:13:08] <deep42thought> ah, right
[15:13:18] <anthraxx> i personally better just do not comment on quarry :P
[15:14:04] <bill-auger> th one i found for perl is 'cpan2aur'
[15:14:20] <abaumann> a, like cpan2rpm, cool :-)
[15:14:43] <bill-auger> most of those language-specific repos do not declare licenses either
[15:15:02] <anthraxx> tooling is great to bootstrap a pkgbuild, but packagers are great to investigate the non obvoisly automation sensitive domain specific optionals
[15:15:54] <abaumann> instead of automatization I personally would rather go into the direction of makepkg helper functions for perl, java, ruby, rust, etc.
[15:16:04] <anthraxx> i believe best is a combination. bootstrap with nice tools but investigate each packages upstream appropriately for such stuff (man, comp, docs, tests etc etc)
[15:16:52] <bill-auger> i looked into several of them - haskell cabal is the only one i found that requires all packages to have freely distributable licenses
[15:17:05] <anthraxx> abaumann: pretty much 0 chance for that, pacman exists with design goal to not become rpm build to debbuild over engineering where you need a PhD in packaging to get what its doing
[15:18:04] <abaumann> ah. that's a point. So, just having good basic documentation for the beginning of an XXX PKGBUILD.. ok
[15:18:18] <anthraxx> yeah or templates etc
[15:19:52] <buildmaster> i686/linux is broken (says nlopc46): https://archlinux32.org
[15:20:33] <deep42thought> errr
[15:20:40] <deep42thought> I'll look into this^
[15:21:13] <buildmaster> pentium4/linux is broken (says nlopc46): https://archlinux32.org
[15:23:43] <abaumann> so dmd and ldc should be ok now.
[15:23:59] <abaumann> I'm still hunting things which could break when we move llvm 10
[15:25:37] <bill-auger> as a side note - a few months back, oaken-source fed all of arch into a parabola autobuilder (network disabled) to find out just how many packages would fail
[15:26:24] <bill-auger> i forgot to ask for a result, or if the experiment ever completed; but it woud be interesting to know
[15:26:38] <abaumann> indeed.
[15:26:38] <deep42thought> yeah, I'm interested, too
[15:26:48] <abaumann> or whether he run out of hardware.. :-)
[15:27:55] <bill-auger> he has something of a po-man's build farm - i think he setup at his school from unused boxes
[15:28:17] <abaumann> cool :-)
[15:42:38] -!- yans has joined #archlinux32
[15:55:28] <deep42thought> for whatever reason, the tarball server did not update its repositories
[15:55:31] <deep42thought> some failing systemd timer
[16:46:11] <abaumann> I start to move clang/llvm stuff now.. and clean up the mess afterwards..
[16:46:52] <abaumann> ..to testing only for now. of course.
[17:04:44] <deep42thought> :-)
[17:05:06] * abaumann is debugging the next generation debugger lldb with the old debugger gdb
[17:24:01] <buildmaster> i486/gdal is broken (says eurobuild6-7-i486): https://archlinux32.org
[18:05:43] -!- torv has quit [Ping timeout: 240 seconds]
[18:06:38] -!- torv has joined #archlinux32
[18:21:15] -!- abaumann has quit [Quit: leaving]
[18:40:14] -!- Funkin-Stoopid has joined #archlinux32
[20:22:33] -!- Funkin-Stoopid has quit [Quit: Konversation terminated!]
[20:37:39] <buildmaster> i686/vibe-d is broken (says nlopc46): https://archlinux32.org
[20:43:20] <buildmaster> i686/appstream-generator is broken (says eurobuild6-6): https://archlinux32.org
[20:45:21] <buildmaster> pentium4/vibe-d is broken (says nlopc46): https://archlinux32.org
[20:46:56] <buildmaster> pentium4/appstream-generator is broken (says eurobuild6-6): https://archlinux32.org
[20:50:52] <buildmaster> i486/redis are broken (says eurobuild6-7-i486) - already flagged out-of-date upstream on 2020-04-30: https://archlinux32.org
[21:07:55] -!- deep42thought has quit [Quit: Leaving.]
[21:12:56] <buildmaster> pentium4/acpi_call is broken (says eurobuild6-5): https://archlinux32.org
[21:32:35] <trotz> 2020/04/30 21:31 CRIT jeti100 Archive Certificate HTTP CRITICAL - Invalid HTTP response received from host on port 443: cURL returned 28 - Timeout was reached
[21:32:35] <trotz> 2020/04/30 21:32 CRIT jeti100 Master Mirror Certificate HTTP CRITICAL - Invalid HTTP response received from host on port 443: cURL returned 28 - Timeout was reached
[21:32:35] <trotz> 2020/04/30 21:32 CRIT jeti100 Master Mirror Secure HTTP CRITICAL - Invalid HTTP response received from host on port 443: cURL returned 28 - Timeout was reached
[21:42:35] <trotz> 2020/04/30 21:42 OK jeti100 Archive Certificate OK - Certificate 'arch.eckner.net' will expire on Fri 12 Jun 2020 02:57:24 AM GMT +0000.
[21:42:35] <trotz> 2020/04/30 21:42 OK jeti100 Master Mirror Certificate OK - Certificate 'arch.eckner.net' will expire on Fri 12 Jun 2020 02:57:24 AM GMT +0000.
[21:42:35] <trotz> 2020/04/30 21:42 OK jeti100 Master Mirror Secure HTTP OK: HTTP/1.1 200 OK - 1408 bytes in 0.081 second response time
[22:16:55] -!- finsternis has quit [Excess Flood]
[22:17:46] -!- finsternis has joined #archlinux32
[22:27:14] -!- send_output-_- has joined #archlinux32
[22:29:29] <buildmaster> i486/broadcom-wl is broken (says nlopc46-i486bs1): https://archlinux32.org
[22:29:31] <send_output-_-> Hello! i am trying to install nvidia-304xx (support was discontinued on september of 2017) on latest arch32. Has anyone tried this?
[22:35:09] <T`aZ> i did tryed
[22:35:22] <T`aZ> and it's painful
[22:35:53] <T`aZ> iirc, you have to tweak the pkgconfig to change the url, to download thd 32bits binary
[22:36:12] <T`aZ> and hope nothing else breaks during compilation
[22:36:43] <T`aZ> i gave up, and i'am using the nouveau driver, it sucks because it's slow, but i'm not gaming on that machine, so i don't really care
[22:37:47] <T`aZ> also, if it works, i suggest to block the updates of the kernel/dkms/and related
[22:38:13] <T`aZ> because the support is obviously lagging, so at each kernel update, things might break
[22:41:00] <T`aZ> because of this, i'm never going to buy nvidia again
[22:44:20] <send_output-_-> thanks for advice!)
[22:45:27] <send_output-_-> in my case, with nouveau i get scary graphical artifacts when trying to use my webcam
[22:45:40] <T`aZ> im' using the same package on another box 64bits and it's working okish
[22:45:57] <T`aZ> yeah, nouveau is really not that good at all
[22:46:14] <T`aZ> (can't blame them honestly)
[22:46:58] <T`aZ> if you want stable/fast/non broken drivers, then spending a bit of time to make nvidia-340xx working is worth it
[22:47:29] <send_output-_-> i m trying to get this shit works 2nd day)
[22:47:47] <T`aZ> yes, be prepared to spend maybe a 3rd or 4th day :p
[22:51:50] <send_output-_-> my blood is already tea a half)
[22:53:32] -!- charims has quit [*.net *.split]
[22:53:32] -!- Alina-malina has quit [*.net *.split]
[22:55:00] -!- charims has joined #archlinux32
[22:55:30] -!- Alina-malina has joined #archlinux32
[22:56:06] <send_output-_-> T`aZ, make tells "no rule to make target"
[22:58:41] <send_output-_-> hah, think i have to downgrade kernel
[22:59:44] <T`aZ> wait, you should get the PKGBUILD, tweak it
[23:00:00] <T`aZ> then makepkg, then pacman -U that pkg
[23:01:28] <send_output-_-> why i cant use installer from nvidia website (sh and binary in one .run file?
[23:02:06] <T`aZ> because it doesn't work out of the box anyumore on recent kernels
[23:02:14] <T`aZ> they stopped to support it
[23:02:59] <T`aZ> so, the idea is to download the official install, apply some patches needed for recent kernels, and then build it, this is done by the aur PKGBUILD
[23:03:49] <T`aZ> if you download the official nvidia binary, there is an option to only extract the "sources"
[23:03:59] <send_output-_-> but this package is no more in AUR
[23:04:32] <T`aZ> then you could apply the patches manually, and then run the official installer from there, without making an official pacman package
[23:05:13] <T`aZ> https://aur.archlinux.org , i'm still seeing it ?
[23:05:13] <phrik> Title: AUR (en) - nvidia-340xx (at aur.archlinux.org)
[23:06:18] <send_output-_-> oh, man all much worse than you think. I mean nvidia-304xx
[23:06:26] <send_output-_-> 304, not 340
[23:06:39] <T`aZ> ha damn, sorry i misread
[23:07:31] <T`aZ> well, hmm, good luck ? :D
[23:07:44] <send_output-_-> i write this from +-10 year notebook
[23:07:51] <T`aZ> the procedure should be the same, get the last official nvidia release
[23:08:00] <T`aZ> apply patches, that now nobody is doing anymore
[23:08:34] <T`aZ> if you take a look at the patches for 340-xx version, there are some chances the needed changes are similar
[23:09:00] <send_output-_-> stop, i am lamer a little. What i can compile (build) if this is proprietary software
[23:09:07] <send_output-_-> ?
[23:09:37] <T`aZ> the nvidia driver has 2 parts, a binary blob, and a layer to interface that binary blob and the linux kernel
[23:09:42] -!- yans has quit [Quit: chaos is the only true answer]
[23:09:43] <T`aZ> that's what you are compîling
[23:10:05] <T`aZ> but as the kernel is always evolving, that layer needs to be adapted also
[23:16:23] <T`aZ> but, are you sure oyu need ? , because from the nvidia page, it seems there are a lot of cards in common between 340 and 304
[23:16:49] <T`aZ> "are you sure you need 304", i meant
[23:20:45] <send_output-_-> yes, my card is GeForce 7400 go
[23:21:39] <send_output-_-> what i will miss if i downgrade kernel?
[23:25:16] <send_output-_-> and stock kernel driver not working with built-in webcam (i use notebook). maybe i should install this one from mxhaard.free.fr/spca5xx.html
[23:25:52] <send_output-_-> my cam is z-star vcc1
[23:26:50] <buildmaster> i486/rime-cantonese is broken (says eurobuild6-7-i486): https://archlinux32.org
[23:29:04] <send_output-_-> and driver from this site made for kernels 2x
[23:29:17] <T`aZ> it depends if you are using new kernel features or not (schedular, drivers etc)
[23:29:36] <T`aZ> as you have old hardware, it's probably not going to impact you to downgrade to an old kernel
[23:30:04] <T`aZ> i would verify systemd though
[23:30:05] <send_output-_-> maybe i should get to the dark side (winXP) ;D
[23:30:33] <T`aZ> well, to be honest, it all depends on your use case :p
[23:31:24] <T`aZ> dont expect to find a recent browser that works under win xp though
[23:31:39] <send_output-_-> it is firefox 52
[23:32:44] <send_output-_-> in all ways, i need winXP for bios update utility
[23:34:49] <T`aZ> may i suggest even an old distro, like debian ? :o
[23:35:09] <T`aZ> maybe that 304 driver is still integrated ? :p
[23:35:11] <send_output-_-> debian 9 supports nvidia-304xx
[23:36:15] <T`aZ> then it's going ot be easier, because rewritting the patched for the 2017 nvidia driver, if you are not a developper
[23:36:19] <T`aZ> is not gonna go well
[23:36:31] <send_output-_-> but issue with webcam is still not resolved
[23:36:33] <T`aZ> patches*
[23:36:49] <T`aZ> yes, you have 2 issues, don't mix them up :p
[23:39:43] <send_output-_-> hmm, developing own driver for unsupported hardware, sound like good activity for next 2 quarantinified months
[23:41:38] <T`aZ> for the nvidia, it is mostly adjusting the interface that changes iwth newer kernel versions, if you look at the the340-xx patches in aur, it's not *that* big
[23:41:53] <T`aZ> no idea bout your webcam
[23:42:16] <T`aZ> buying an 40€ usb and supported one is probably easier
[23:42:56] <send_output-_-> so if i still need winXP and my bios supports booting only from usb-fdd, can you advice program that can do this
[23:43:15] <send_output-_-> bootable usb-fdd*
[23:44:37] <T`aZ> im not sure i understand your question
[23:47:40] <T`aZ> copying the win xp iso to a usb flashdrive *might* work
[23:49:08] <send_output-_-> bios doesnt support usb-hdd (default modern bootable mode for modern pendrives, so i need to format my flash disk drive as diskette (usb-fdd), so that BIOS can boot from it
[23:49:58] <send_output-_-> usb-fdd or usb-zip*
[23:51:32] <send_output-_-> i dont know exactly which of these two supports bios
[23:52:19] <send_output-_-> in bios there feature called "boot from external FDD"
[23:54:14] <send_output-_-> i heard that whether the flash drive works in USB-FDD mode depends on which chip is used