Tuesday, 2023-03-14

*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC00:07
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc00:33
*** gnucode <gnucode!~gnucode@user/jab> has quit IRC01:49
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC02:26
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc03:05
*** tplaten <tplaten!~tplaten@> has quit IRC03:20
*** tplaten <tplaten!~tplaten@> has joined #libre-soc03:34
*** tplaten <tplaten!~tplaten@> has quit IRC03:51
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC04:26
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc04:53
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC05:11
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc05:45
ghostmansd[m]TL;DR: GitHub started revoking write access to open source from guys who happen to work in Russian tech companies.06:01
programmerjakewell, what I saw in some phoronix comments made me think otherwise (though I didn't check if they were right), apparently that person's employer is sanctioned by the US govt. so GitHub may not have had a choice. GitHub hasn't done anything to many other Russian developers, so I'd guess this is not a GitHub-out-to-get-all-Russians scenario.06:04
programmerjakeif that's the case, I'd guess that's the US govt. over-reacting...06:07
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC06:15
sadoon[m]In any case, host your own git.06:19
sadoon[m]And mirror to convenient services06:19
programmerjakeor at least mirror to more than one hosting service06:19
sadoon[m]Preferably not github lmao06:19
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc06:47
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC07:39
programmerjakefor meeting times, note that north america started DST already, europe is starting in a week or so, and australia is ending in two weeks or so.08:23
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc09:02
*** josuah <josuah!~irc@> has quit IRC09:27
*** josuah <josuah!~irc@> has joined #libre-soc09:29
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC09:50
lkclprogrammerjake, markos, if you have discussions somewhere please ensure they are cross-referenced into the bugreport.  i have no idea what you were talking about and cannot review it10:36
lkclregardless 4 instructions vs almost 30 is absolutely no contest whatsoever https://bugs.libre-soc.org/show_bug.cgi?id=1015#c610:36
markoswe had the discussions here, I think programmerjake actually linked to the irc discussion10:37
lkclno, there is no link in the bugreport.10:37
lkclthe job of the IBM Architects is to allocate opcode space10:37
lkclif they have 30 instructions to allocate they're going to get pissed off10:38
lkcl(IBM has the ultimate authority and control *over-and-above* the OpenPOWER Foundation!!)10:38
lkclthis is not generally well-known10:38
markosanyway, as I said to me it meant more consistent with the existing naming scheme, but I do not reject the new scheme10:38
markosand yes 4 vs 30 is a good argument10:39
lkclif however we request only 4 instructions (sotto voice: with some extra mode-bits that are elsewhere, not in the XO Field) their job is made vastly simpler and they are less likely to complain10:39
markoswhat are the chances of them actually proposing the old naming scheme themselves?10:40
markosok then10:40
lkclbecause IBM as it currently stands is still focussed on VSX10:40
lkcland they don't really spend that much focus on the Scalar ISA10:40
markoswell, I beg to differ there, they seem to have totally dropped the towel on it and don't really give a shit at all about anything, but anyway10:41
lkclso it's not so much that they wouldn't propose the *naming* scheme, but more that they wouldn't propose these kinds of scalar instructions *at all*10:41
lkcl[because VSX equivalents already exist, IBM has VSX, so why would they consider putting redundant instructions in?]10:41
markosbecause they also consider VSX dead but that's another story10:42
lkclit's a latency thing.10:42
lkclyes, the propagation of information is slow, and opinions differ.  one of the very interesting facets of massive companies that people forget, there is not "one homogenised IBM/Microsoft/Apple/etc/etc opinion"10:43
markoswhich is a good thing, but I would really love for a change to see a new fresh opinion, and in practice10:44
markosI really do have high hopes that SVP64 will change things for the platform otherwise Power is going the way of the dodo and will just another system in my retro/vintage collection :)10:45
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc10:48
markostoshywoshy, btw, OT, I ended up installing gitea from source on a Talos vm, installed without an issue, the snap was full of problems10:50
markoswhen done I will start importing repos and installing CI and see how it goes, I might ask you a few questions in the process :)10:50
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC10:58
sadoon[m]<markos> "toshywoshy, btw, OT, I ended..." <- Awesome that you confirmed that, I definitely will use gitea or forgejo on my POWER810:59
sadoon[m]I wonder how problematic nextcloud will be though11:00
markosI installed it on debian 12, golang 1911:00
sadoon[m]And I plan to use debian/devuan instead of my current ubuntu setup too11:00
sadoon[m]Ah, a beta tester it seems11:00
sadoon[m]They froze already though right?11:00
markosonly use a couple of ubuntu vms because of a couple of snaps11:00
markosdebian you mean11:01
sadoon[m]Might be time for me to start working on sffs deb12 then perhaps11:01
sadoon[m]markos: Yeah11:01
sadoon[m]In fact wasn't the freeze literally this Sunday o_o11:01
markoswell it's almost frozen and I always do that, I migrate to testing a little before release11:01
markosand then stick with the released version for a long time11:02
* sadoon[m] is tempted to do the same, soon..11:02
markosso far no problems on ppc64le at least11:02
sadoon[m]I swear though, if I find no good way to treat debian like a source-based distro I will actually write up a full program to do just that11:03
markosonly hardware I'm actually missing right now, surprising as it is, is arm11:03
markosonly cloud VMs which suck11:03
markosbecause I cannot just spin vms as I like, that would cost a ton of money11:03
sadoon[m]I remember in fosdem me and octavius: met a guy who works on open source PCB designs11:04
sadoon[m]He had a 16-core ARM "desktop"11:04
markossadoon[m], what do you want to do? have an autobuilder for debian packages?11:04
sadoon[m]We got his business card just in case11:04
markoswell, there is an ampere workstation, with up to 128cores11:05
sadoon[m]markos: Yes and one that can work similar to portage with no fuss11:05
markosbut it's not that fast11:05
markosand it's expensive11:05
markosI was thinking a mac studio11:05
markosbut that's also terribly expensive and not upgradeable11:05
markosthough it's insanely fast11:05
sadoon[m]And running linux is still alpha stage on those11:06
markosso about the autobuilder, do you want to replicate what debian is doing?11:06
markosor just add a few packages in a private repo and autobuild them?11:07
sadoon[m]Only sort of11:07
sadoon[m]What I want is to be able to bootstrap it all from source11:07
sadoon[m]With full dependency resolution11:07
markosif you want the former then you need to replicate the debian infrastructure11:07
markoscbmuser on #debian-ports might be of help11:07
markoshe's doing that on a regular basis for multiple non-release architectures11:08
sadoon[m]Huh, definitely know him/her11:08
markosyeah he's basically the maintainer of every port that is non-release11:08
sadoon[m]I remember that handle, I'll have a chat there then11:08
sadoon[m]Later today I guess11:08
markosdefinitely him, John Paul Adrian Glaubitz11:09
sadoon[m]You can tell I'm tired from work since I referred him/her after reading you say "he" :)11:09
markoshe even managed to get m68k support back in11:09
markosand alpha and sparc, etc11:09
markosprogrammerjake, lkcl just a thought, could there be just the 4 instructions and the other be aliases? that way no additional opcodes are needed11:57
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc12:03
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC12:10
sadoon[m]We might be in luck12:27
sadoon[m]Apparently apt-build still works?12:27
sadoon[m]It automates repo creation as well12:27
markosin the past I used mini-buildd for package building the reprepro for merging own repo with official debian -so that the customer would see a single debian + own overlay repo in their systems12:33
markoss/the reprepro/and reprepro/12:33
sadoon[m]Only problem is it doesn't do any dependency tree resolution a la emerge12:35
markosthat's why I used mini-buildd :)12:36
sadoon[m]For example, it would rebuild packages one by one with the assumption that the dependencies already exist12:36
sadoon[m]markos: But it looks like a pain to set up :(12:36
markosoriginally tried OBS (Open Build System) by SUSE but that proved way too problematic12:36
markosOBS would enter a loop building the same packages again and again and again12:37
markossadoon[m], it's not as hard, but I think it's not really maintained anymore12:38
sadoon[m]It isn't afaict12:38
sadoon[m]I looked into it about a year or so ago12:38
markosah no, it's in Debian even!12:38
markosjust got in unstable12:38
markosseems quite maintained12:39
markosworth a look12:39
sadoon[m]Yeah worth it12:40
sadoon[m]Nothing to lose really12:40
sadoon[m]Not like we have many options xD12:40
lkclmarkos, the aliases were what i suggested 6+ months ago.13:24
lkclmodes-in-a-table creates a standard-RISC-paradigm uniformity13:24
markosso, there is no problem really then?13:26
sadoon[m]Got mini-buildd setup on sid13:45
sadoon[m]This is gonna need a lot of coffee13:45
sadoon[m]Probably wise to utilize the openpower server we have once I've got the hang of this, then things can become much more organized13:46
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc14:05
sadoon[m]From what I understood so far, it would be better to use debian buster as the mini-buildd server since it has a stable release of it. It's going to build all the packages in a chroot anyways so it doesn't make much of a difference what host os it is14:08
markosyes, there is no need to get the latest unless you need particular features14:14
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC14:24
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc14:54
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC15:27
lkclsadoon[m], all the tools that you need *will* be there - it's just that the debian developers do not go out of their way to "promote themselves" if you know what i mean15:35
lkclmany distros go out of their way "look at us! look at us! we're better than any other distro because we have reproducible builds! we're the first evva!!!!! give us money!!!!"15:36
lkcland it turns out that actually the only reason that distro could do the work so quickly is because TEN YEARS PREVIOUSLY, debian and fedora did all the hard work in the low-level infrastructure to become the *actual* first distros to do reproducible-builds15:36
lkclbut then didn't shout loudly about it15:37
sadoon[m]They kinda built their reputation and don't need all the hype, they know their stuff is good15:37
lkclconsequently you _will_ find all the tools you need to build entirely from source - but you'll have to do a little digging and/or asking around15:37
lkcland they just get on with it.15:37
lkclthe wiki is more an "aide-memoire" just like ours is15:38
lkclso if you get onto oftc.net debian IRC channels you'll find they'll be pretty helpful15:38
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc16:44
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC17:53
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc17:54
markoslkcl, just occured to me, only the assembler file needs to be compiled with the power assember18:52
markosall the rest of the files could be compiled using normal C so you wouldn't have to include all those python dependencies18:53
markosbut in that case you cannot link the ppc64le .o file directly18:53
markosin that case you need the .bin file to load straight in the simulator18:53
markosI can modify the code to make it run on x86 as well18:54
markosand I will do that after I finish chacha20, and for all the examples, I would have to build an x86 libresoc dev environment though to make sure they work18:54
markosso people might choose a ppc64le or x86 dev environment (or whatever for that matter as long as there is a ppc64le (cross) assembler available18:55
lkclmarkos, yep, correct: and the binutils-ppc64-le cross-compiler needs to be compiled and installed from source anyway with the svp64 support19:05
markosgood, I'll tackle that after I'm done with chacha20, which shouldn't take that much longer19:07
*** gnucode <gnucode!~gnucode@user/jab> has joined #libre-soc19:52
*** lxo <lxo!~lxo@gateway/tor-sasl/lxo> has quit IRC19:58
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC20:11
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc21:13
*** gnucode <gnucode!~gnucode@user/jab> has quit IRC21:13
*** gnucode <gnucode!~gnucode@user/jab> has joined #libre-soc21:14
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has quit IRC23:21
*** ghostmansd[m] <ghostmansd[m]!~ghostmans@broadband-109-173-83-100.ip.moscow.rt.ru> has joined #libre-soc23:52

Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!