Tuesday, 2022-05-03

*** kylel1 is now known as kylel07:10
lkclprogrammerjake, i added whitespace-recognition to the page-reader for pseudocode so that comments with indentation are correctly ignored10:53
programmerjakelkcl: i'm going to sleep now, feel free to revert once you got the fix to work, the commit you pushed didn't fix it.11:15
lkclprogrammerjake, ok willdo11:19
lkcldidn't realise comment-checking was in so many places. am slightly on autopilot11:26
ghostmansd[m]Hi folks, JFYI, from now on, I'll use my private email, so I've subscribed to our archive from it11:40
ghostmansd[m]I've also sent my public key, but this message is pending, because I'm not member yet. Perhaps idiotic Gmail's dot in email might confuse the mailing archive.11:41
ghostmansd[m]My address I use and post is ghostmansd in Gmail, but when I initially created it, it was ghostman.sd. These are the same, Gmail doesn't care about dots.11:43
lkclghostmansd[m], will take a look11:46
lkclmailing lists care :)11:47
lkcli've added that address as "approved to post" so you should be fine now11:48
ghostmansd[m]Hm, somehow I didn't receive this message on local email, only on 3mdeb...12:04
ghostmansd[m]Ah, got it, I need to confirm12:05
ghostmansd[m]lkcl, could you, please, add the public key to git and talos1?12:53
lkclyep 1 sec..14:14
lkclghostmansd, sorted14:17
ghostmansd[m]lkcl, star14:19
ghostmansd[m]Ok looks like git works with new pubkey17:47
ghostmansd[m]lkcl, I've just checked sv_binutils.py, it works after your changes :-)17:50
ghostmansd[m]I see a strange commit claiming to clean whitespaces up, though17:50
lkcldeep joy17:50
lkclah yes. 80 char limit17:50
ghostmansd[m]But, instead of cleaning those...17:50
ghostmansd[m]It adds few in the end of the line17:50
ghostmansd[m]I'm kinda confused17:51
ghostmansd[m]at least terminal highlights this17:51
lkclyes that'll be my mistake17:51
ghostmansd[m]BTW, as for line break limit17:51
ghostmansd[m]Perhaps we could establish some sort of git hook?17:51
lkclunfortunately in some cases (URLs in comments) it's actually necessary17:52
ghostmansd[m]Ah, OK, fair enough17:52
lkclstackexchange for example17:52
lkcli'll do the whitespace now17:53
ghostmansd[m]Missed your message17:59
ghostmansd[m]I did the same thing, but a bit more...17:59
lkcldoh :)18:00
ghostmansd[m]Ok, done18:00
lkcllooks good18:01
lkcli'm kinda of the generally-vague opinion that if it looks clean / tidy it's easier to understand18:02
ghostmansd[m]Same :-)18:02
lkclbut it may just be the fact of making changes that causes "learning"18:02
lkclhey if it works18:02
ghostmansd[m]I prefer spacing twice on function declarations, though18:02
ghostmansd[m]And always indent by fixed amount, never giving a crap about padding to method name18:02
ghostmansd[m]In my experience, function names and methods change names and arguments two often to consider them for a padding18:03
lkcli do wish autopep8 wasn't borked when it comes to function arguments18:04
ghostmansd[m]But I've never given a f*ck to this on code reviewa18:04
lkclf(x,y,z,a,b,c,d,e,f,g,h,i,j...) is correctly formatted by autopep818:05
lkclas is f1(...) and f12(...)18:05
lkclbut the moment you go above 4-character indentation on the function-name-plus-bracket it goes to hell18:05
lkclreally weird bug18:05
lkclyou have to actually like... have a good reason18:06
ghostmansdactually PEP8 is not the strongest parts of Python18:06
ghostmansdmany dubious choices, like "hey, two line breaks between functions, but not inside the class"18:06
lkcland "the lead developer has to have about 10 xterms side-by-side due to memory recall problems" is a pretty big one18:06
ghostmansdI mean, why the heck these are different?18:06
lkclas is "for developers with less resources i.e. low-resolution screens a large width is a problem"18:07
lkcloh, yes.18:07
ghostmansdlkcl, I've updated the email on binutils commit, it was surprisingly easy...18:08
lkcli rationalised that one as, "if it's top-level (no indentation) then 2 lines is good"18:08
ghostmansdand this will be the same email which I used to contact FSF earlier18:08
lkclahh brilliant18:08
lkclyeah ~/.gitconfig or something?18:08
ghostmansd> [if it's top-level] Well, I'm actually interested how they treat when you have nested functions :-)18:09
ghostmansd~/.gitconfig or something?git rebase -r <some commit before all of your bad commits> --exec "git commit --amend --no-edit --reset-author"18:09
ghostmansdTaken as is from stack overflow, can I call myself wise now, eh?18:09
ghostmansdactually kinda neat, I liked the overall idea18:10
lkclyyeah i'd kiiinda prefer not to have a full commit-history rewrite?18:10
lkclbecause there are some places using git submodules and i'm not sure if the rewrite would cause merry havoc there18:10
ghostmansdit for sure can :-)18:10
ghostmansddepends on how badly you deal with it, though18:11
ghostmansdas any force rewrite18:11
lkcllike, "whoops that commit listed in the .gitmodules file no longer exists"18:11
ghostmansdthough, generally, this might be the problem with 3rd party modules18:11
ghostmansdactually I'm unsure of what to do with openpower-isa18:12
ghostmansdI'm thinking of leaving it as is, frankly18:12
ghostmansdbecause, whilst binutils are easy (nobody depends on them)18:12
ghostmansdopenpower-isa is used by many people, and I don't want to break anyone's work18:12
ghostmansdalso, binutils DO have linear history18:13
ghostmansdand sv_binutils.py in openpower-isa doesn't18:13
lkclyeah it's pretty critical, it's the simulator, wiki overlay, and hardware, all depend on it18:13
lkclhave to be extremely careful18:13
ghostmansdwhat'd you recommend? at this stage, I'd like to keep any published stuff to be published w/o affairs to external companies18:14
lkclpfhhh... ok go for it18:15
ghostmansdand, frankly, I'm not sure what to do here: if I publish binutils patch with a couple of files which are clearly auto-generated, then anyone can go and check the affiliation18:15
lkclwell copyright assignment there should take care of tha18:16
ghostmansdyou mean FSF copyright assignment, right?18:16
ghostmansdbut this doesn't concern openpower-isa repository... or does it?18:17
ghostmansdbecause, well, we have the generated files in binutils...18:17
ghostmansdbut sv_binutils.py resides at openpower-isa18:17
ghostmansdfrankly I'm noob in licensing, so I have no idea18:18
lkclwell as the output from a compiler (which is what sv_binutils.py is), copyright decisions can be made on the *output* from the compiler which have absolutely nothing to do with the copyright of the *tool*.18:18
ghostmansdyeah, and we intend to provide the output to binutils-gdb18:19
ghostmansdI mean, well, the compiler is also FOSS18:19
lkcland that's where it gets interesting, going forward, because we *may* have to assign the copyright of openpower-isa to the *OpenPOWER* Foundation18:19
lkclit is, after all, trademarked18:19
ghostmansdaha, this really gets interesting18:19
lkclit'll be part of the submission of what we're doing to the OPF ISA WG18:20
ghostmansdso, yes, if we assign sv_binutils.py copyrights to FSF, this would really deal badly with OpenPOWER18:20
ghostmansdI mean, this would be a real conflict, right?18:20
lkclah not quite18:21
lkclbecause part of the deal with the FSF is a "re-licensing back"18:21
lkclbut yes it's a... weird situation :)18:21
ghostmansdOK, I guess the current approach is OK: we grant the copyright to output18:22
ghostmansdThat is, nothing but binutils-gdb files participate18:23
ghostmansdso, OK, there's only one reason to change the committer email now :-)18:23
lkcllxo, wasn't there something about that in recent discussions? or was that for gcc?18:24
ghostmansdbut for me even that one is sufficient, so I'm going to do it now18:24
lkclyep go for it.18:24
lkclwe'll sort out any mess afterwards18:24
ghostmansdthank you!18:24
* lkcl salutes :)18:24
lxoone issue is copyright, the other is whether the generated files qualify as source code.  is one expected to modify those generated files in the process of developing binutils, or would one rather use the input to the tool?  in the latter case, the input is the source code, and its output is object code even if it looks like sources18:30
lxo(under the terms and understanding of the GPL)18:30
lkclthe generated files are definitely source code: they're sufficiently regular - and so large - that it would be quite insane to write them by hand18:32
lxothat's really not what matters.  the relevant question (per convo with rms not long ago, on an unrelated issue) is which one you'd rather modify in the course of development18:34
ghostmansdHm. In this particular case, the generator itself.18:35
lkclyou _could_ hand-edit the autogenerated file if you really wanted to18:36
ghostmansdI mean, I can do it either way... but the correct one is affecting the generator.18:36
lxolkcl, it's not about what could be conceivably possible, it's about what one would rather do, which form would be most convenient to modify the program18:49
lxosay, if the generator were to be lost forever, one could then decide that the generated file, being sufficiently modifiable, is good enough to carry on the development.  but otherwise, the generator's inputs (and possibly the generator itself, depending on details) would probably be it.18:50
lkclit's just that the inputs are strictly speaking - or have to become - the property of the OpenPOWER Foundation.18:53
lxoas for copyright assignments, if the generator is conceived of as a compiler, its inputs and the outputs are really just different forms of the same work.  I don't think (but IANAL) one could assign copyrights over the latter but not the former18:54
lkclbeing, as they would become, part of the OpenPOWER ISA Specification18:54
lkclwhat might turn out to be the saving grace is that specifications themselves are not copyrightable18:56
lkcl(but the document *containing* the specification *can* be)18:56
lxo*nod*.  it's tricky :-(19:00
lxoanyway, I have to go now.  'later19:00
lkclok :)19:01
ghostmansdOK, it seems like I finally came up with something that seems to work19:36
ghostmansdThis URL https://pastebin.com/v4x7dF7v contains the script19:36
ghostmansdThe overall idea is to create a new commit with author reset, then override the original one19:36
ghostmansdFolks, before I even consider pushing it with $(git push --force-with-lease) or even simple $(git push --force), I'd like to have as many eyes as possible on this. And, ideally, if we can, I'd like to have yet another backup of openpower-isa repository.19:38
ghostmansdlkcl ^19:38
tplatenI'm trying to find out which type of DDR3 module the orangecrab has19:44
tplatenIn ls2.py I find from gram.modules import MT41K256M16, MT41K64M1619:45
programmerjakerewriting history will have bad effects such as reversions losing what commit was reverted because the sha1 changes, as well as the many links to specific commits from bugzilla, emails, the wiki, and more...is there any way you can add your old email as an alias of your new email in a .mailmap and have that be sufficient ghostmansd?19:45
programmerjakesee https://github.com/llvm/llvm-project/blob/main/.mailmap for an example19:46
tplatenI had a look at coldboot.c, it copies the flash content to dram, which I currently do not have working on the orangecrab yet19:47
ghostmansd[m]programmerjake, I'm rather worrying about the whole affiliation with foreign company.19:51
programmerjakehmm...i'll let you decide but wanted to mention that there will be bad sideeffects of rewriting history, also it'll still be pretty easy to figure out your previous affiliation if one wanted to...19:53
ghostmansd[m]And, unfortunately, some of commits were done after these embargoes were applied, and after weapon was transferred to Ukraine.19:54
ghostmansd[m]I do not really bother about the fact that affiliation had placw. :-)19:55
ghostmansd[m]I bother about the fact that any new submissions, including such big as binutils, might raise unneeded discussion and attract unneeded attention.19:55
ghostmansd[m]This is, I want these binutils works, and future ones, to be done by me outside of any companies. I simply do not want to take this risk.19:56
ghostmansd[m]Ergo, before submitting anything such public as binutils, I'd like to leave this controversy out of scope.19:57
programmerjakeso imho if there are any legal problems, they won't change by rewriting history.19:57
programmerjakeok. would it work to have the previous history in a history tag or something so links to commits don't turn into 404s?19:58
ghostmansd[m]I think yes, why not.19:59
ghostmansd[m]On the other hand, I think we might leave openpower-isa out of it.19:59
programmerjakeok, in that case, go ahead and rewrite. maybe have the tag have a message stating it corresponds to commit xyz that is the rewritten version20:00
ghostmansd[m]Ok, let's wait for Luke20:01
programmerjakesounds good!20:01
ghostmansd[m]On the other hand, I think that the whole discussion depends on how FSF treats sv_binutils.py.20:06
tplatenI had a look at https://orangecrab-fpga.github.io/orangecrab-hardware/r0.2/OrangeCrab-r0.2-ibom.html20:09
programmerjakehmm...iirc gcc dropped the copyright assignment requirement...did binutils do that too?20:10
ghostmansd[m]As for the sources and code contributed to binutils, I clearly want this to be committed by me personally w/o any references to foreign company. I have no idea what laws will be here by the time I complete these works. Considering how the situation changes, I want to make things as simple as they can be.20:10
tplatenU4 is Micron Technology Inc.MT41K64M16TW-107:J TRMT41K64M16TW-107_J-TRBGA-96_9.0x13.0mm_Layout2x3x16_P0.8mm120:10
programmerjaketplaten: neat website, didn't know they added that feature to kicad20:11
programmerjakeghostmansd[m]: yeah, makes sense20:13
lkclworks for me20:15
tplatenI knew that Neo900 UG had developed some kicad addon, I don't know if it is still actively maintained.20:22
tplatenThe software package is called eeshow.20:23
tplatenlast commit is 5 years old20:25
ghostmansd[m]After giving it a thought, I think programmerjake is right. I'll simply create a .mailmap file, and that's it for openpower-isa.20:40
programmerjakei'll join the meeting in a few min...21:56
lkcljn, lxo sadoon_albader[m markos meeting22:06
ghostmansd[m]A nice summary of contributing to binutils22:25
*** psydroid is now known as psydroid[m]23:42
*** psydroid[m] is now known as psydroid23:43

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