IRCloggy #git 2006-11-08

Logs Search ←Prev date Next date→ Channels Documentation

Provider of IRC logs since 2005.
WARNING: As Freenode became unjoinable and lost all warnings in topics, we cannot log channels on Freenode anymore.

2006-11-08

dwmw2_zzz → dwmw2_gone00:08
Insount → Insount_away00:28
Insount_away → Insount00:43
Insount → Insount_away00:51
dwmw2_gone → dwmw2_TPE01:02
→ kanru joined01:28
→ anholt joined01:41
→ GeertB joined01:50
beu → boo02:33
boo → beu02:43
→ benlau joined03:00
→ sgrimm_ joined03:07
→ sgrimm joined03:33
→ yashi joined03:49
cehteh http://www.pipapo.org/gitweb?p=cinelerra-svn;a=summary how could master/origin lag behind the actual head? that repo is only synced from svn with git-svnimport, but i recently has some problem and maybe i did a improper fix04:13
mugwump that's a gitweb bug I think04:21
cehteh a way to fix it?04:23
mugwump it's a cosmetic bug, so I didn't worry about it04:27
→ sgrimm joined04:28
cehteh ok .. well looks scary on a first view ;)04:28
→ spearce joined04:58
→ Newsome joined05:03
→ gitster joined05:10
gitster is cehteh still around?05:13
cehteh yes05:13
gitster the cinelerra problem you mentioned about an hour ago.05:13
that is a problem at the repository side.05:14
it's info/refs is out of sync.05:14
cehteh ok05:14
how can i bring it back into sync?05:14
gitster "git-update-server-info".05:14
cehteh too easy :P05:14
mugwump aha05:15
mugwump updates his server refs05:15
cehteh done :)05:15
thanks05:15
gitster Whenever a ref is updated (usually a public repo that is pushed into this is handled via post-update hook) you need to run it to support http downloaders.05:15
cehteh i just add that to my cron entry05:15
gitster If the site runs more recent version of gitweb, it is not needed to have gitweb behave, but http "commit walkers" still need it up-to-date.05:16
what do you use to autosync with svn?05:16
cehteh its debian/etc05:16
h05:16
git-svnimport -rm -A authors -T trunk/hvirtual svn://svn.skolelinux.org/cinelerra/05:17
well a week ago it broke with some debian update .. some libs where in a wrong version ..05:18
gitster Ah, I do not know if git-svnimport has post update hooks, but if you are driving that from your cron job twice a day, running update-server-info from that cron job after svnimport returns would be the right thing to do.05:18
cehteh i likely fixed that improperly05:18
yeah .. i just add that to the crontab entry now05:18
gitster sounds good.05:18
cehteh planning to make a better script someday which also repacks .. but thats not urgent05:19
btw .. did anyone noticed the sync-timeout patch i send to the ML? :)05:21
gitster yes, and I did not quite like it. first, I think driving "prune" from cronjob is, how would I say it..., playing with fire?05:22
cehteh repack calls prune05:23
gitster by the majorly lockless nature of git, prune is unsafe with anything that modifies the repository; local update-index, commit, fetch, etc.05:23
cehteh yes05:23
gitster if I were doing it I would probably take the repository offline, run prune and then put it online again, but I probably would not do that from a cron job or make the job interval sufficiently long.05:24
I do not think "give up and do not remove if we cannot sync in time" is necessarily a bad idea,05:25
cehteh i am planning to do only incremental packing .. if someone merges/pulls during a prune wouldnt it just abort and he has to retry?05:25
gitster but I do not think that timeout belongs in the prune (or in git).05:25
cehteh gitster: i first wanted a mail saying that this is a bandaid05:26
actually calling sync() in prune is already a bandaid05:26
gitster Why? I would say it is a prudent thing to do. That is where we really would want to make sure the previous writes hit the platter.05:27
Oh the other hand,05:27
cehteh i know that it is not the best solution .. but i think its at least an improvement to the current situation where a prune can hang for long time05:27
gitster even if previous writes hit the platter (we could use fsync or fdatasync if we want to make it cheaper), there is no guarantee that the renaming of *.lock file to the final name hits the platter without sync X-<.05:28
cehteh yes .. bit ugly05:29
i have no better solution either05:29
gitster I would really think the "sync taking a long time" is a local matter. Why can't you do an equivalent of ^C from surrounding script you run from cron which runs whatever command (one of which happens to be "git prune")?05:29
cehteh in that case it wasnt even from a cron script05:30
i just repacked the repo and it hung05:30
on the console .. and you cant ^C the sync .. it will hang05:30
maybe you can argue that it was a bad idea to zero out a 80GB hd in another console ..05:31
but having git hanging because of some totally unrelated job which is going on is not a good idea i would say05:32
besides its not really intuitive why it hangs for a user05:32
gitster if you try not to talk about sync needing to write the whole dirty data which may take a long time, then that is true -- it is hard to explain -- but you have exactly the same problem of being hard-to-explain if you do the backgrounding of the sync and erroring it out. Your users would demand to know why it does not finish and errors out.05:38
cehteh mhm .. maybe a suggestion "try later again.." would leave a better feel for the user05:44
← avuton left05:53
→ robinr joined06:37
gitster somehow feels #git is active mostly according to european time-of-day. when he drops by, which is not that often, not many people are around...06:45
cehteh well .. 7:45am here06:46
(germany)06:46
gitster 2245 tuesday (california)06:46
cehteh but i always feel that most channels are active at us time anyways ... evening/early night06:46
european morning and us nighttime is a bad time to expect traffic on irc ;)06:47
do you feel bored or what?06:48
gitster not really, just feeling left out somewhat ;-) #git feels more european than others. many active people seem to be from europe.06:49
cehteh here wasnt that much traffic the past days anyways06:51
→ kanru joined06:54
cehteh git stores no (extended) metadata about versioned files aside? (execpt permissions)06:58
→ ShadeHawk joined07:05
→ cods joined07:15
→ ferdy joined07:56
→ cods joined08:04
→ lu_zero joined08:12
lu_zero hi08:13
← gitster left08:37
→ nud joined09:02
→ devogon joined09:08
→ kanru joined09:29
→ chris2 joined10:20
→ wildfire joined10:36
wildfire with cogito, how do I checkout the contents of the tree as of a particular tag?10:36
ahh - cg-seek10:37
→ yash1 joined10:47
→ matled joined10:51
→ robfitz joined11:34
→ GyrosGeier joined12:05
→ cods joined12:19
→ kanru joined12:42
→ kanru_ joined13:24
Insount_away → Insount13:26
→ cods joined13:30
→ GeertB joined13:48
dwmw2_TPE → dwmw2_gone13:58
→ xl0 joined14:02
dwmw2_gone → dwmw2_TPE14:38
→ timlarson_ joined14:39
→ timlarson_ joined14:49
→ timlarson__ joined15:12
→ Gitzilla joined15:30
→ GeertB joined15:43
→ GeertB joined16:17
→ cods joined16:28
→ krh joined16:34
dwmw2_TPE → dwmw2_zzz16:49
→ robinr joined16:57
pasky hmrz16:57
repo 1931 0.0 0.0 2864 1364 ? S 15:51 0:00 /bin/sh /home/pasky/bin/git-fetch -f -u -k --mirror-all git://git.kernel.org/pub/scm/linux/kernel/git/jsipek/gq.git16:57
fetch from k.org hangs forever again :(16:57
→ cworth_ joined17:05
robinr EGIT preview: http://rosenberg.homelinux.net/paste/lp4.png :)17:06
does git use different kinds of compression? I had to stop reading the history because java's compression routines bombed17:08
→ spearce joined17:13
cworth_ → cworth17:13
robinr spearce: http://rosenberg.homelinux.net/paste/lp4.png17:14
spearce robinr: VERY COOL. That's slick. 17:14
pasky (a hint: please be careful about showing full 40-digit hashes in UI, 7, 8, or 12 digits suffice at most places)17:17
spearce yea, i agree with pasky; those object ids can probably be abbreviated to 8 digits; just to make it more readable.17:18
spearce reminds himself to add OBJ_OFS to pack parser in jgit.17:19
robinr well, it is probably possible to do more. This is the first version that works at all :)17:20
spearce i was terrified of writing the compare interface for eclipse. i do some eclipse/swt hacking at my day job and its ugly... :)17:21
robinr it is synchronous, slow as hell (probably not my fault :) ) and there are many considerations on exactly how the history should be presented17:22
you want to know a secret?17:22
or, I can just take credit for it all :)17:23
I didn't write a compare interface, I just interfaced with the compare interface17:23
spearce so they already had most of that ui built?17:23
robinr the history list is mine17:24
it is the full history until reading crashes..17:25
spearce until reading crashes? is there a bug in jgit that i don't know about?17:25
robinr java.util.zip.DataFormatException: unknown compression method17:26
spearce aww crap.17:26
→ matled joined17:28
spearce is that the egit repository itself you are reading back through? packed?17:29
robinr egit itself, packed17:30
spearce i'll try to setup a test case here and cause that... maybe i can fix it. :)17:30
robinr commit id: 84a899697aad8ed660db4605b5c8d0b30c18d6f517:31
tree: 6ad2df1c74434d7df8d76fbf6678e36352c4609917:32
→ alley_cat joined17:34
→ Eludias joined17:53
→ luxgladius joined18:01
luxgladius Hey there, does anybody know how to turn off less pagination where it is default?18:05
Gitzilla Pipe the output through cat.18:08
spearce cworth: PAGER= git foo 18:08
odd. I wrote 'or: ' and bitchx wrote 'cworth: '.18:09
luxgladius Ok, thanks for the tips18:09
pasky spearce: use irssi ;)18:14
ShadeHawk do any of you have on-line papers or presentations about git (for example for OLS), which could be put in http://git.or.cz/gitwiki/GitLinks ?18:14
spearce starts installing irssi...18:15
pasky ShadeHawk: http://pasky.or.cz/ols-cogito/ for me, Junio has it on the web as well, it went through the mailinglist shortly after the presentation18:16
and jdl has it on web as well, I guess18:16
Gitzilla Gitster made his slides available and the proceedings are available as PDFs. Check the ML archives for the slides link.18:16
→ sgrimm joined18:21
ShadeHawk Thanks18:24
→ lyakh joined18:25
cworth spearce: I don't know what it is about IRC clients but many of them have totally insane nick completion.18:27
spearce pasky: how do i get irssi to do my freenode nick password? 18:28
→ spearce_ joined18:33
spuk- spearce: use nickserv.pl, or use /network -autosendcmd "/quote nickserv identify pazzw0rd"18:33
spearce_ → spearce18:33
spearce robinr: I think I have offset delta implemented in egit now. I have to run off to lab but I'll look at the uncompression crashing. Git uses straight zlib (which Java uses the same library I think) so this is more likely just us reading the wrong bytes (bad offset seeked to and read).18:53
→ spearce joined19:00
→ segher joined19:27
→ Gitzilla joined19:28
→ anholt joined19:50
→ anholt_ joined19:50
Insount → Insount_away19:54
→ DrNick joined20:17
spearce thanks the git gods for verify-pack -v.20:22
spearce Apparently jgit isn't reading a pack properly; its figuring that a deltified tree is using a blob as its base, which then makes it a blob. That can't be right.20:22
ShadeHawk is it even possible to use other type element as a base?20:27
→ Newsome joined20:31
spearce ShadeHawk: No, the file format doesn't permit it. There's a bug in jgit that is grabbing the wrong base. :-)20:40
ShadeHawk Ah20:41
spearce I keep trying to walk through it in a debugger but these students in my lab section keep interrupting me asking for assitance. :)20:41
robinr give them an interesting problem to solve instead :)20:41
and you can help them with it.20:42
spearce they are parsing xml in php. *shudders*20:42
ShadeHawk is it really that hard?20:50
"pure" PHP or can they use SimpleXML for example?20:51
By the way, speaking of XML, what do you think about idea of moving from RSS 2.01 to Atom for gitweb "feeds"?20:52
robinr Does it make things better?20:55
spearce: hmm, reading now works... :)20:55
it seems the use of BufferedReader is the problem.20:56
I removed one reader and set the buffer size to 1 in another place.20:56
it seems the buffered tried to read past the end of an embedded stream ... or something along those lines20:58
spearce ShadeHawk: They were trying to use any php library they could to read an RSS stream. But its a course for non-programmers so these folks even have trouble making simple HTML pages and the like.20:59
robinr: I just found a delta base resolution that's horribly broken. Its not the bug you originally reported and its not related to the buffer problem you are talking about now.20:59
ShadeHawk from what I have read Atom feeds can have consistently and easily fragments of HTML/XML as contents21:00
spearce but its keeping me from reading back a tree. :-(21:00
robinr the same tree I reported, or another?21:00
spearce ShadeHawk: maybe, but their assignment was a trival: get text and print it. :)21:00
robinr: same tree.21:00
different pack though so it may have a different base, etc. then what you have.21:01
robinr I'm trying to send you a patch via DCC right now21:01
spearce yea, i doubt that will work. email (spearce@spearce.org)21:02
robinr git needs and IRC interface21:05
I don't suggest you apply the patch, it's more of a "something like this"21:06
ShadeHawk well, there is work on GitTorrent protocol, why not IRC interface (you mean IRC channel of transfer, weren't you?)21:07
pasky that's not practical since the data is rather heavily ratelimited21:07
robinr DCC too?21:07
pasky aiui dcc doesn't work for the people who would make use of it21:07
you can pass patches around with dcc rather easily too21:08
robinr well I couldn't in this case :(21:14
→ gitster joined21:17
gitster pasky here?21:20
→ sgrimm joined21:21
pasky yeah21:21
gitster did I break the fork support?21:21
pasky sick but here21:21
gitster ah, what did you catch?21:21
pasky you are talking about the next rebasing?21:21
gitster No, I have a few gitweb patch in master last night to "fix" breakage for users who do not want to use project forks in gitweb.21:22
pasky it was a flu since last week to today... since today it transformed to enteric flu or something, so I'm feeling really great21:22
hmm21:22
I'll update repo.or.cz and see what happens then21:22
gitster: still seems to work well21:27
even w/ latest next21:27
gitster thanks. I tested with both configurations.21:27
robinr spearce: why the ugly placement of braces in egit? :/21:31
gitster by the way, I think somebody said delta in pack for tree is never created based on a blob and the restriction is in the file format. That is a misinformation.21:32
So the reader of a pack should be prepared to see that; existing pack-objects never try to find delta between different types, but that is just an implementation issue.21:34
→ spearce_ joined22:07
spearce_ gitster: then tell me how to get the type of a deltafied object without assuming it from the base. :)22:08
pasky ShadeHawk: is atom and rss compatible?22:10
spearce_ robinr: as far as brace placement goes, that's how i like 'em. :) since i wrote the code i chose. of course now that someone else is also hacking on it with me maybe the sun java standard or a more gittish standard might be better.22:13
nud pasky: what do you mean by compatible ?22:13
gitster A delta has n-byte type and length (4-bit type, and 7-bit-per-byte length), followed by 20-byte base object name, followed by delta data, so the type is already there in the header.22:13
nud they share principle but have nothing common wrt syntax22:13
gitster But you may be right. Recent delta-base-offset change may have broken that premise -- which is not a big loss.22:13
spearce_ go reread that code. :) the type is 6 or 7 to mean ofs_delta or ref_delta.22:13
that's not recent change either; that's been since deltas were introduced.22:14
nud but syntax-wise, rss is not compatible with itself when you have two distinct versions :p22:14
pasky hmm22:14
then I think ATOM shouldn't _replace_ RSS22:14
nud atom is more rational afaik22:14
better conceived22:14
robinr is it supported by a large majority of RSS readers?22:15
mugwump if an RSS reader supports all RSS versions, it will probably support atom too22:16
nud afaik google only use atom for their webservices22:16
gitster shawn, you are right. c62266f (2005-06-30) made it so.22:17
ShadeHawk well, I'll _add_ Atom then (after some uniquifying). I hope some time this decade ;-))))22:17
spearce_ and its annoying its that way too because you have to walk back the entire chain just to obtain type.22:18
spearce_ → spearce22:18
ShadeHawk Atom has author and contributor which naturally map into git's author and committer I think. I don't know enough about RSS 2.022:18
gitster yeah, we are trading density with seeks.22:19
sgrimm Any git-svn folks in the house?22:19
spearce probably not anything majorly wrong with that as we usually don't go get the type of an object unless we are also getting its content.22:19
ShadeHawk (by the way, there were request on IRC about adding RSS feeds for branches (heads), which wasn't unfortunately followed by code even thoug the suggestion how to do was posted here)22:20
robinr knewsticket didn't understand it at least22:20
spearce jgit on the other hand does some retarted object type checking early before getting the data, so its seeking 2x. :-(22:20
but that's just poor coding on my part; not a fault of the file.22:20
pasky ShadeHawk: dunno, well, it's really so trivial that... :)22:20
ShadeHawk spearce: it's egit, jgit?22:20
robinr ok, that's where som of jgits speed comes from22:21
spearce jgit is the pure java git library; egit is the eclipse wrappers on top of that.22:21
ShadeHawk ah22:21
spearce i wanted jgit to not be based on eclipse at all so other folks (e.g. ant) could use it too.22:21
robinr: yea. :-)22:21
→ DrNick joined22:23
spearce dammnit. when i seek within a pack file in jgit i'm going to the right location but find the wrong data: (176 != 89).22:32
ShadeHawk pasky: I started coding this, and it is not _that_ trivial (URL, description, using href subroutine)22:39
this = rss for branches22:40
pasky hmm22:51
→ Oejet joined23:09
pasky gitster: BTW, can we please get git-pickaxe renamed to git-blame now?23:14
(sorry if it already was, I'm not watching the list clsoely now)23:14
spearce nah, use git-knife. :)23:14
pasky still has his disagreeing response to git-pickaxe's -M / -C use postponed too :(23:15
Oejet Since when did archaeologists use pickaxes?23:17
spearce git-dust-brush23:18
robinr git-accuse23:19
spearce git-convict23:20
robinr git-sue23:20
why not make it a flag to git-annotate?23:20
spearce hmmm. git-sue might make it more friendly for the corporate types.23:20
sgrimm git-pillory?23:21
robinr hmm, have to look that work up23:21
git-audit23:21
that's the corporate version23:22
pasky complete git solution for your law and accounting departments23:22
spearce there's big money to be made there. :)23:23
Oejet The corporate command would be: git show-copyright-infringements23:24
gitster the command internal is already prepared to pretend to be compatible with annotate/blame so we can make three synonyms whenever we want. My original plan was to rename it to blame before pushing it out, but Linus posted quite a big advertisement on the list and mercurial (or somebody else, I do not remember) user survey mentions request for something like "git-pickaxe", so maybe the name needs to stay.23:24
robinr no23:27
Oejet is looking a word beginning with 'o' in the Linux kernel, so that the command could be abbreviated to "git-sco", but to much dismay fails to find any such word.23:27
Oejet *for a23:27
robinr Oejet: maybe if pickaxe could locate copied code from outside git23:28
gitster oejet, since nethack ;-)23:29
robinr git-sco could be git-grep. like git grep -l SMP|mail court23:30
or whatever they used23:30
← gitster left23:30
spearce needs git-find-and-fix-bug23:30
Oejet Yeah, or what if you could turn your repository into a mobile agent (the fashion of the day), so that it would _by itself_ go out into the world looking for infringing code. Clever, eh?23:31
pasky spearce: we call the first part git bisect ;)23:36
spearce pasky: only if it ever worked to begin with. :)23:36
robinr apply it to itself then23:38
sgrimm Anyone know of any frontends to git-svn to give it a more svn-like UI? Trying to convince a team to migrate to git and if I can reduce the amount of unfamiliarity it'll be an easier sell.23:38
robinr the whole point of git is that it is different23:38
pasky spearce: oh well :)23:39
robinr svn-like ui = ?23:39
tortoisesvn?23:39
pasky the common practice is that when you want to write a tool, you start from scratch with an empty file, write few functions, then start adding commands...23:40
perhaps it's time for a more revolutionary appraoch23:40
write and maintain a monster that does _everything_23:40
and when you want to write a tool, you start with that and keep removing stuff23:40
sgrimm robinr: Like a one-step "update" command that does the sort of thing cg-update does (merge upstream changes into the working copy). Or a "commit" command that refuses to let you blow away changes in the svn repository if you haven't fetched the latest.23:40
pasky is full of helpful ideas today23:40
sgrimm I am ready to write a frontend to git-svn, just trying to see if I'd be reinventing the wheel.23:41
Gitzilla pasky: you go first23:41
robinr sgrimm: how stuck are people into the "SVN" model?23:42
sgrimm Unfortunately given the political reality here, we will have to do a gradual transition from svn to git, and I'd like to make it as seamless as possible to reduce pushback.23:42
Oejet is not good at politics.23:42
sgrimm robinr: It is much more convenient than the git model for a subset of developers (e.g. making one-off edits to HTML files.)23:42
robinr stay with svn for a while and convert a few frontrunners using git-svn as-is23:43
sgrimm robinr: That's what I'm doing now. I'm one of the frontrunners. But it is too easy to shoot yourself in the foot with the current git-svn UI; you have to be very careful to do everything just right or you lose changes.23:43
robinr sgrimm: if the SCN model suits you better, maybe git isn't for you23:44
pasky sgrimm: you can configure git/cogito to behave exactly according to the svn model23:44
including update-before-commit23:44
spearce sgrimm: i'm in the same position as you, only not with svn, with a f'ing system whose only interface is through Internet Explorer. Try keeping ~10,000 files in sync with git without shooting yourself in the foot.23:45
sgrimm pasky: Yes, but the problem is that I can't use Cogito to check in to our svn repository, I have to use git-svn for that.23:45
robinr I don't commit to any SVN repo, however I do the same with CVS23:45
sgrimm And, for example, if I do git-svn dcommit and fail to notice that one of the files I'm committing has just been changed by someone else, git-svn will happily overwrite the other version with my version without so much as a warning.23:45
spearce oh that's baaaad.23:46
sgrimm (I personally feel that's a bug, or at least a design flaw, in git-svn, but I'm too new to it to say that with any authority.)23:46
spearce at least i can lock the file and compare contents before overwriting it.23:46
robinr but I keep my patches in stgit, which simplifies things when they get back throug cvsimport23:46
my patches become empty and I can delete them. Until then they that in my repo23:47
the same model should work with git-svn.23:48
sgrimm Basically my problem is that we have a mix of people using svn right now. A bunch of developers who will adapt to git just fine and will end up preferring it, and a bunch of less technically inclined folks (web designers, etc.) who will never have any need for git's advanced capabilities and for whom the svn model is perfectly adequate. I want to come up with something that'll work for both groups during a transition period.23:48
robinr however it seems you are looking for a centralized model, and git ain't so23:48
with stgit+git-svn your local changes won't be overwritten "just like that"23:49
sgrimm robinr: It's not my local changes that are getting overwritten, it's the changes in the svn repository.23:50
robinr ah, that's bad23:50
sgrimm Yeah. :)23:50
robinr that's different from git-cvscommit23:50
it applies patches23:50
it's clearly a git-svn bug23:51
sgrimm It actually surprised me a lot to see the behavior when I was first testing git-svn. I did an svn commit from a plain svn client, changed the same file on the git side, and expected to get at least a warning message when I tried to commit from git-svn. Instead it just sent the git version to the svn repository and when I did an update from the svn client, the svn-side change was gone.23:51
robinr bring it up on the mailing list23:52
pasky sgrimm: you're a no-windows shop?23:52
sgrimm pasky: Yeah, pretty much. The Windows guys will have to use git-cvsserver, but I think they don't use their workstations as more than glorified X terminals anyway. Most developers here use Macs.23:52
pasky how much work it would be to code git-svnserver?23:52
nice23:53
spearce sgrimm: give the nontechnical folks git-gui.23:53
pasky give them qgit, it's nicer and stuff ;)23:53
pasky hides23:53
sgrimm haha23:53
spearce or qit.23:54
pasky a competition!23:54
that's always good23:54
robinr a problem with nontechnical (and some technical) is that merge conflicts scares the hell out of them23:54
sgrimm I will write up the git-svn problem and send it to the mailing list. I'm glad I'm not the only one who sees it as a bug!23:54
robinr that's my some SCM tools have file locking23:54
s/my/why/23:54
spearce ah, good point. git == no locks. i'm fighting that right now too with some users.23:55
robinr svn doesn't lock either23:55
sgrimm robinr: Happily, we have enough conflicts in svn that everyone is okay with them, even the less technical people (though they sometimes have to ask for help.)23:55
robinr locks are evil23:55
spearce i agree they are evil but try explaining that to a user who edits the same word document as 90 other people, and they only communicate by writing crap in that file. :)23:56
sgrimm So that's not a barrier. Actually the biggest barrier to going 100% git will be the changes in workflow, but cogito simplifies stuff enough that it probably won't be a big deal. ("cg-update" is very convenient.)23:56
robinr qgit is nice too23:57
and graphical :)23:57
spearce note to self: add graphics to git-gui. :)23:57
robinr :)23:58
not to self: go to bed23:58
pasky sgrimm: (cg-commit --push might be interesting for people deep set in the svn workflow)23:58
sgrimm pasky: Yeah, that will definitely be very nice for people who want to ease their way into the git world.23:59
spearce starts to wonder if this jgit bug is actually the jvm unable to seek within a file... but it claims that it did...23:59
Oejet Git _is_ harder, because it is more general, and thus more _abstract_.23:59

Logs Search ←Prev date Next date→ Channels Documentation