IRC logs for #openrisc Tuesday, 2012-08-21

stekernjuliusb: it's up to you, if you have some good plan how to organize them, I'd say merge them now05:23
stekernif not, just push it, I can take a look at merging it in with main later on then05:25
juliusbstekern: ah I've just merged them11:35
juliusbno problems11:35
juliusbI'll make the default "fourstage" pipeline your cappuccino one11:36
juliusbnow that there's a 'smaller' pipeline in the espresso one, there's no point having the fourstage guy11:36
stekernok, cool11:37
stekernthere are some caveats still in cappuccino though, you can't disable icache for example11:39
juliusboh hehe11:41
juliusbtoo bad, use espresso then11:41
juliusbmmmm espresso11:41
juliusbit's that time :)11:41
stekernyeah, and eventually you'll be able to disable it, so it's just a phase11:42
juliusbi'm curious about rdiez's new core in orbuild11:43
juliusbi'm wondering how he verifies it11:43
juliusbi like the idea of a systemc testbench11:43
juliusbultimately is more powerful than Verilog11:43
juliusbas a verification environment11:44
stekernisn't he "just" running these?
juliusbOK, not sure.11:47
juliusbI assumed he was also running against the GNU testsuite?11:47
juliusbcan you give me a bit of guidance on some git stuff11:48
juliusbi've got a local branch called "doc" which I want to become the master on the github11:48
juliusbi've got the github in there11:48
juliusbas a remote, called github11:48
juliusbcan I just go: git push -u origin:master doc11:49
juliusband that'll rename the doc branch as master?11:49
juliusbyou can do something like this if I recall11:49
stekerngit push doc origin:master11:49
stekernI think11:50
juliusbor i should just update my local master to the doc head i think11:50
juliusbthat's probably easier11:50
juliusbyeah i think that's done it better11:51
stekernno, it's git push origin doc:master11:52
stekernyeah, I try to keep my local branches named the same as the remote ones, I'd just mix everything up otherwise11:53
juliusband it's live: push11:55
juliusbshit, no licenses on the files11:56
juliusbi completely forgot about that11:56
juliusbsnap decision to be taken11:56
juliusbwhat do you reckon for license?11:56
stekernheh, you pick one ;)11:58
juliusbi want my one11:58
juliusbI can always change it later11:58
stekernso use yours then :)12:00
stekernbut very nice that it finally has surfaced from the dark corners of your harddrive =)12:12
juliusbi know12:14
juliusbsorry it took like 3 months12:14
stekernah, well, what do they say, better late than never12:16
stekernI'll steer my attention back to it as soon as I've got this toolchain/compiler business out of my system ;)12:17
juliusbOK licenses updated12:26
juliusbdidn't change your i/d cache files12:26
juliusbI'll add you with commit privileges on that account actually12:26
stekernwould it be a bad idea to start an openrisc organisation on github?12:31
juliusbnot sure what that means12:32
juliusbgotta run off, bbl12:32
stekernwell, essentially it's the same as an github account12:33
stekernlook at milkymist as an example:
stekernjust as a way to tie together all the people around using github, like me, you, rdiez, pgavin, franck, antmicro etc12:37
jeremybennettstekern: An organization is a good idea. It's a shame that Marcus won't set up OpenCores git repo, because it is when we move from individual to collective effort that maintaining the OpenCores community becomes important.12:39
jeremybennettSomething to discuss when we meet next month. There is some admin overhead to running an organization on GitHub.12:40
jeremybennettWe use one for the Synopsys ARC Linux and GNU tool chains.12:40
jeremybennettstekern: Will you set it up on GitHub?12:41
stekernsure, I can do that if it's something we want12:42
jeremybennettWell, in the absence of anything from OpenCores it makes sense.12:42
jeremybennettThe downside is that it again splits the OpenCores community, because we'll end up with an organization Wiki, mailing list, etc, etc on GitHub.12:43
jeremybennettYou start to wonder what is the point of OpenCores!12:43
jeremybennettIIRC you can control which members of the organization have write access to which repos.12:46
stekernyes, I'm a member of the milkymist organisation, but I only have write access to clang and llvm repos12:46
jeremybennettThat will be useful for us, so we can manage who maintains different components.12:46
jeremybennettWill you make it an OpenRISC organization or an OpenCores organization. Probably makes sense to do the former.12:47
jeremybennettI'll move Or1ksim there - I've asked for months to do that on OpenCores with no luck!12:48
jeremybennettback later12:49
stekernok, openrisc organization added, now I just need to figure out how this works ;)12:56
stekernI put up me, jeremybennett and juliusb as owners for now13:02
juliusbis there some link?13:09
juliusboh yeah, just
stekernI choose lower case for the name to avoid having the address as
juliusball good13:10
juliusbhow do I list the mor1kx guy on the organisation?13:13
stekernas I understand it, there are two options: 1. you can just go to your page and press the fork button and choose openrisc. 2. you can transfer the repo control to openrisc, but then it will disappear from your repo list13:14
juliusbahh OK13:14
juliusbso I have to fork it?13:14
stekernultimately, I like option 1 better, since if I would transfer control to openrisc, I'd probably would like to keep a personal fork where I do more experimental hacking13:15
jeremybennettstekern: Excellent - I see it there13:16
stekernI mean, you can still push to the forked openrisc one at the same time you push to your own if you wish13:16
jeremybennettjuliusb: you might want to transfer your repo to openrisc and then fork back. Semantically it is exactly the same thing, but it means that the repo won't show in openrisc as a fork from your personal repo.13:19
stekernyeah, true13:19
juliusbthat sounds fine13:19
juliusbbut what are the rules, everyone who's an owner of the openrisc group can commit to anything repo there?13:20
stekernyes, but you can create "teams" with more fine-grained permissions13:20
juliusbwell for now I'm happy to have anyone there capable of committing to it13:21
jeremybennettNew or1ksim repo created. Now I just need to add the code...13:25
stekerngiven that you have a commit history from svn for it, it might be a good idea to import it with git svn and then push it13:27
jeremybennettOh yes - I know how, it just needs the time to actually do it! Watch this space.13:28
jeremybennettArrggh - now my local SVN is version 1.7 and I need to run svn upgrade over the entire tree!13:32
jeremybennettI may be some time...13:32
stekernhuh? can't you just do: git svn clone
juliusbstekern: so I have something I want to push to both my newly forked mor1kx repo and the normal one (a README.pod file)13:43
juliusball I have done is cloned my newly forked repo13:43
juliusbhow do I push to both? do I have to do 2 individual things?13:43
stekernand then git add remote <github/openrisc/or1ksim> and push to that?13:43
stekernthat was for jeremybennett ;)13:44
stekernI don't know if you can simultaneously push to 2 places in the same command, but add a seperate remote and then do 2 pushes13:45
juliusbyep cool13:45
stekernseems to be a guide how to save you from having to do 2 pushes13:46
juliusbah, already did the first13:46
juliusbOK cool13:49
juliusbman github is really nice!13:57
juliusbi never really knew everything it had13:57
juliusba wiki, graphs13:57
juliusbability to annotate commits13:57
stekernheh, yeah. I hadn't noticed the graphs earlier14:03
jeremybennettstekern: that is a good point - using the source repo. It's a bit more complex because of the structure of trunk and branches, but it's the right idea.14:04
stekernlooking at mor1kx, I have deleted more lines than I have added ;)14:04
juliusbi've just realised mor1kx-defines has a heap of implementation `defines which aren't used14:07
juliusbso i'll delete them to avoid confusion14:07
juliusbthey're all parameters now14:07
jeremybennettstekern: Just tried the following:14:07
jeremybennettgit svn clone --trunk= --branches= --tags=
jeremybennettI got an immediate failure:14:08
jeremybennettUse of uninitialized value $url in pattern match (m//) at /usr/libexec/git-core/git-svn line 491.14:08
jeremybennettUse of uninitialized value $_[0] in substitution (s///) at /usr/share/perl5/File/ line 341.14:08
jeremybennettfileparse(): need a valid pathname at /usr/libexec/git-core/git-svn line 49714:08
simoncookdo you need to give it a directory to clone into?14:09
jeremybennettIt should clone into the current directory if not specified. However I think I have misunderstood the meaning of --trunk etc.14:09
simoncookis it init instead of clone?14:09
jeremybennettclone should be the equivalent of git svn init followed by git svn fetch14:10
stekernI've never tried getting the branches and tags, so you're on your own there ;)14:10
jeremybennettJust checked - I haven't misunderstood, those do seem to be good args.14:10
jeremybennettWell I really want at least the tags because we have been assiduous in tagging releases!14:11
jeremybennettUnfortunately this has been my general experience of git svn. Something breaks in the PERL for anything but an absolutely standard SVN repo layout.14:12
stekernI read up on, I don't think git-svn will do what you want like that14:16
stekernthe svn2git seems to do what you want, create local branches and tags from the svn branches and tags14:17
jeremybennettAh yes - I think I've been here before.14:17
stekerngit-svn will set-up references to the svn14:17
jeremybennettThe problem seems to be with the svn git --trunk arg - it doesn't seem to work as advertised.14:18
simoncookjeremybennett, ive done a search, try using git svn init instead, then edit the .git/config to remove // replacing with / in the branches and tags, then try git svn fetch, that at least is pulling svn commits14:25
jeremybennettsimoncook: Thanks for that.14:26
jeremybennettLunch is now calling loud and clear!14:26
stekernjeremybennett: something like this should work: svn2git -v --trunk trunk/or1ksim --tags tags/or1ksim --branches branches/or1ksim14:36
stekernmake sure to make a or1ksim directory and cd into it first...14:37
stekernI didn't and it made a mess in my ~/openrisc/ directory :(14:37
stekernyou might want to use an authors.txt file too to map the users14:44
stekernsvn2git -v --trunk trunk/or1ksim --tags tags/or1ksim --branches branches/or1ksim --authors ../authors.txt14:56
stekernwith authors.txt being something like this:
stekernjeremybennett: since I already did all this, I can spare you the trouble, you can clone from this temp repo :) git:// 15:04
jeremybennettstekern: Thanks - I'll pick that up15:24
jeremybennettI ran svn2git here, but it didn't get the tags15:27
stekernstrange, I got them15:30
jeremybennettstekern: My mistake. No use using git br -a to list the tags. They did all come over. I've cloned your temporary repo.16:14
jeremybennettJust tidying up stuff now.16:14
jeremybennetto1ksim now fully on GitHub with all the SVN history, tags and branches. Thanks to stekern and simoncook for assistance.17:07
juliusbGreat news!17:20
jeremybennettI suppose at some time I ought to roll in the outstanding patches and make a new release17:27
juliusbstekern: have changed all references to fourstage to cappuccino20:33
stekernjuliusb: nice one! I thought about doing that, but obviously you beat me to it =)20:54
_franck_cappuccino, espresso, cool names :)22:00

Generated by 2.15.2 by Marius Gedminas - find it at!