March 28, 2015 
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | >
March 28, 2015 [00:36:54] *** brendan has quit IRC[01:07:43] *** finkVC has quit IRC[01:09:37] *** amacks has quit IRC[01:10:31] *** brendan has joined #fink[01:26:42] *** asmeurer has joined #fink[01:34:46] *** mischi has quit IRC[02:00:14] *** asmeurer has quit IRC[02:03:08] *** asmeurer_ has joined #fink[02:04:08] *** howarth has joined #fink[02:04:25] <howarth> any one understand this error in clang[02:04:27] <howarth> for[02:04:28] <howarth> sprintf(level_text,(level == -1)?"?":"%d",level);[02:04:45] <howarth> warning: data argument not used by format string [-Wformat-extra-args][02:05:09] <howarth> /usr/include/secure/_stdio.h:47:56: note: expanded from macro 'sprintf'[02:05:09] <howarth> __builtin___sprintf_chk (str, 0, __darwin_obsz(str), __VA_ARGS__)[02:05:39] <howarth> with the carets on level and __VA_ARGS__[02:08:10] *** asmeurer_ has quit IRC[02:09:43] <howarth> this sprintf has the comment[02:09:47] <howarth> /* if level is unknown, use "?" */[02:09:49] <howarth> above it[02:12:16] <howarth> https://developer.apple.com/library/prerelease/ios/documentation/Cocoa/Conceptual/Predicates/Articles/pSyntax.html[02:12:33] <howarth> suggests perhaps darwin doesn't accept the ? mark[02:41:15] *** howarth has quit IRC[02:53:41] *** howarth has joined #fink[03:04:39] *** finkVC has joined #fink[03:04:40] <finkVC> jwhowarth: dists/10.7/stable/main/finkinfo/utils expect-10.9.patch[03:04:40] <finkVC> remove dead patch[03:05:16] <finkVC> jwhowarth: dists/10.7/stable/main/finkinfo/utils expect.patch[03:05:18] <finkVC> add debian/fedora patches and others[03:05:44] <finkVC> jwhowarth: dists/10.7/stable/main/finkinfo/utils expect.info[03:05:46] <finkVC> use updated patch[03:06:09] <finkVC> jwhowarth: dists/10.7/stable/main/finkinfo/utils expect-10.9.info[03:06:10] <finkVC> use updated patch[03:06:37] <finkVC> jwhowarth: dists/10.7/stable/main/finkinfo/utils expect-10.10.info[03:06:39] <finkVC> use updated patch[03:21:07] <finkVC> jwhowarth: dists/10.7/stable/main/finkinfo/utils expect.info[03:21:09] <finkVC> perl edit away one more warning[03:21:24] <finkVC> jwhowarth: dists/10.7/stable/main/finkinfo/utils expect-10.9.info[03:21:26] <finkVC> perl edit away one more warning[03:21:40] <finkVC> jwhowarth: dists/10.7/stable/main/finkinfo/utils expect-10.10.info[03:21:42] <finkVC> perl edit away one more warning[05:10:38] <finkVC> babayoshihiko: dists/10.7/stable/main/finkinfo/sci qgis26-py.info[05:10:40] <finkVC> Add missing BuildDep[07:46:58] *** jack has quit IRC[07:54:24] *** jack has joined #fink[07:54:48] *** jack is now known as Guest19468[08:10:42] *** finkVC has quit IRC[08:26:00] *** badshaah has joined #fink[08:33:41] *** Guest19468 is now known as jack[08:33:51] *** jack has quit IRC[08:33:51] *** jack has joined #fink[08:43:12] *** asmeurer has joined #fink[08:53:00] *** asmeurer has quit IRC[09:15:12] *** badshaah has quit IRC[09:15:43] *** badshaah has joined #fink[09:20:08] *** badshaah has quit IRC[09:31:45] *** finkVC has joined #fink[09:31:45] <finkVC> babayoshihiko: (3 files in 1 dir)[09:31:45] <finkVC> New upstream version as a new package[10:10:26] *** u-ha has quit IRC[10:11:23] *** badshaah has joined #fink[10:35:46] *** howarth has quit IRC[10:36:10] *** howarth has joined #fink[10:40:16] *** badshaah has quit IRC[11:24:50] *** mischi has joined #fink[11:26:01] *** RangerRick has quit IRC[11:26:20] *** RangerRick has joined #fink[11:46:41] *** badshaah has joined #fink[12:18:23] *** badshaah has quit IRC[12:31:44] *** finkVC has quit IRC[15:57:57] *** mischi has quit IRC[17:39:10] *** mischi has joined #fink[18:00:17] <howarth> any git experts around?[18:00:41] <howarth> what is the best way to roll back a git checkout to the file state at a specific commit[18:11:44] *** mischi has quit IRC[18:57:08] <dak180> howarth: `git reset --hard <rev-id>` is likely what you want[18:57:29] <howarth> thats what I'm doing[18:57:39] <howarth> the dejagnu git is a mess[18:58:09] <howarth> it has a master, a dejagnu-1.5 and dejagnu-1.5.2 branch but no 1.5.1[18:58:39] <howarth> they regressed the gcc test suite summary results in the 1.5.2 release[18:58:49] <dak180> howarth: be careful though that will cause the commits you have orphaned to eventually evaporate (in a week or two with default settings)[18:59:04] <howarth> I can get the same working behavior in the dejagnu-1.5 branch at its final commit[18:59:15] <howarth> http://git.savannah.gnu.org/cgit/dejagnu.git/commit/?h=dejagnu-1.5&id=52dad3fd75c64d8e8cc0de5f8ca123c4aa7c32d2[18:59:50] <howarth> but I can't seem to get any position in master or dejagnu-1.5.2 branch to bahave like 1.5.1[19:00:24] *** asmeurer has joined #fink[19:00:50] <dak180> howarth: depending on what you are trying to do it may be better to just make a new branch at the point you want checked out and leave the existing branches where they are[19:01:07] <howarth> I am trying to regression hunt[19:01:14] <howarth> 1.5.1 works and 1.5.2 is broken[19:01:31] <howarth> I am trying to find the change which introduced the breakage[19:01:55] <howarth> assuming their branches work like svn, and additional fixes can be added[19:02:22] <dak180> howarth: in that case i suggest the `git bisect` command though i recommend reading the docs before using it[19:02:29] <howarth> I am starting to suspect that once 1.5 branched (and was used up to the 1.5.1 release) some fix was added that never made it back into master[19:03:09] <howarth> the bit that really confuses me is that I can't seem to roll back master to not show the bug no matter how far back I hard reset[19:05:07] <howarth> http://lists.gnu.org/archive/html/dejagnu-commit/2011-03/msg00007.html[19:05:24] <howarth> Sun Mar 6 21:11:38 2011 +1100 was the creation of the 1.5 branch[19:06:08] <dak180> howarth: bisect should give you the answers you are looking for[19:10:41] *** asmeurer has quit IRC[19:10:47] <howarth> I am confused by the differences between the commit, the tree and the parent[19:11:20] <howarth> for example[19:11:21] <howarth> http://git.savannah.gnu.org/cgit/dejagnu.git/commit/?id=014ae81e6009eebec2975d29b5a7deb69694632a[19:12:12] <howarth> if I have pulled[19:12:43] <howarth> git clone git://git.sv.gnu.org/dejagnu.git[19:12:52] <howarth> it defaults to master right?[19:13:23] <howarth> I was only able to get the 1.5.1 behavior back by using[19:13:39] <howarth> git reset --hard 6bd8dddc81d8a5f2e835f148a9be1509f3edb147[19:13:51] <dak180> howarth: the commit is the id of that commit the tree is an internal git object that you will almost never need to worry about in practice and the parent is the commit id(s) of the commit(s) that came before the current one[19:13:57] <howarth> which is the parent commit for http://git.savannah.gnu.org/cgit/dejagnu.git/commit/?id=52dad3fd75c64d8e8cc0de5f8ca123c4aa7c32d2[19:14:50] <howarth> I can double check but I am pretty sure for that change if I used the commit rather than the parent I didn't get the desired bahavior[19:15:24] <howarth> dejagnu behaved broken as in 1.5.2[19:15:43] <howarth> so again[19:15:44] <howarth> for http://git.savannah.gnu.org/cgit/dejagnu.git/commit/?id=52dad3fd75c64d8e8cc0de5f8ca123c4aa7c32d2[19:15:47] <dak180> howarth: (keep in mind two things: everything in git is internally identified by a sha1 hash and commits are a directed graph)[19:15:53] <howarth> which is the final commit in the 1.5 branch[19:16:02] <howarth> what would you expect to need to do[19:16:08] <howarth> from a clean tree[19:16:09] <howarth> I had to[19:16:47] <howarth> git clone git://git.sv.gnu.org/dejagnu.git[19:16:47] <howarth> cd dejagnu[19:16:48] <howarth> git reset --hard 6bd8dddc81d8a5f2e835f148a9be1509f3edb147[19:19:26] <dak180> howarth: if i wanted to find the commit that broke something from an initial clone of a git repo and i knew both a point where it had worked and where it did not i would make note of those two commits (or tags) and run `git bisect` after having reviewed the documentation on the command so i could be sure of the correct invocation[19:20:42] <dak180> howarth: `git reset` is not the command for that exercise unless you really want to go bug hunting manually ☺️[19:20:46] <howarth> one thing that confuses me is how the branches are handled[19:21:10] <howarth> in svn you have to do a separate pull but git bundles it all together no?[19:21:30] <howarth> so after cloning how to you switch back and forth between the branches and master[19:21:38] <dak180> howarth: ja, with git you get the whole repo[19:22:06] <howarth> what I am suspecting is that 1.5 and master forked after the branch[19:22:28] <howarth> and by they used 1.5 branch to develop 1.5.2[19:22:33] <howarth> err[19:22:35] <howarth> 1.5.1[19:22:59] <howarth> and by the time they did 1.5.2 they branched directly out of master rather than 1.5 branch[19:23:18] <howarth> ok this is a bit better[19:23:28] <howarth> git reset --hard 52dad3fd75c64d8e8cc0de5f8ca123c4aa7c32d2[19:23:40] <dak180> howarth: the first thing i would suggest is a gui since that is the only way i have been able to keep such things straight[19:23:48] <howarth> which was the commit for http://git.savannah.gnu.org/cgit/dejagnu.git/commit/?id=52dad3fd75c64d8e8cc0de5f8ca123c4aa7c32d2 works like 1.5.1[19:24:30] <dak180> howarth: i recommend using http://gitx.github.io/[19:24:38] <howarth> my gut instinct is that a fix got applied to 1.5 branch that never made it back to master[19:27:56] <howarth> in which case, I probably have to roll all the way back to the 1.5 branch creation and walk forward in master[19:29:19] <dak180> howarth: part of the reason i recommend a gui is that there are three states you have to keep in mind for any given branch: what is actually on the server, what is in your local clone's remote for that server (what you local copy know that the server had the last time it talked to it) and finally your local branch which is what you actually commit to and otherwise move around[19:30:53] <dak180> howarth: but as i said what you want is the http://git-scm.com/docs/git-bisect command; it will find the exact commit that introduced your problem for you[19:31:35] <howarth> after testing the 1.5 release commit[19:31:43] <howarth> I plan on walking from http://git.savannah.gnu.org/cgit/dejagnu.git/commit/?id=67e77e4589db843d7cc9683bb7ddbf7c6ca3b1dd[19:31:56] <howarth> which is * NEWS: Start a new section for post-1.5 changes.[19:32:18] <dak180> howarth: and if you have a test script already setup you can use it with the bisect command so it can find the problem commit unattended[19:32:34] <howarth> I should be able to find a commit in master where the failures begin[19:35:16] <dak180> howarth: my point is that if you are aware of even only one commit each that you know is bad and good bisect will find the commit that boke things fairly quickly; usually in 6 steps or less in my experience (usually less)[19:36:37] <howarth> we don't know that[19:36:48] <howarth> we only know that 1.5.1 works but 1.5.2 doesn't[19:37:12] <dak180> howarth: that is all it needs to know[19:37:14] <howarth> the problem is 1.5.2 came out of 1.5.1 branch[19:37:49] <howarth> so the issue could actually date back prior to 1.5.1[19:38:05] <howarth> and had been something fixed in 1.5.1 branch that never got fixed in master[19:38:22] <dak180> howarth: that they are on different branches means nothing to its ability to find the issue[19:38:57] <dak180> so long as they have a common ancestor[19:39:19] <howarth> but aren't branches roughly the same as in svn[19:39:39] <howarth> in that once you branch changes can be made to the branch which aren't mirrored in master[19:40:05] <howarth> my point is that I have to reghunt much further back[19:40:27] <howarth> since they reused 1.5 branch rather than creating a 1.5.1 branch[19:40:51] <dak180> howarth: my point is that the bisect command is designed to exactly that[19:44:21] <dak180> howarth: just from a quick glance at the repo it seems like `git bisect start dejagnu-1.5.2 dejagnu-1.5.1-release` would get the ball rolling where you want to go[19:44:54] <howarth> I am making progress now with a manual walk[19:45:09] <howarth> I just need to move through master[19:45:20] <howarth> okay http://git.savannah.gnu.org/cgit/dejagnu.git/commit/?id=2354ff747ff0bda5a8c18d7554b92afa812f6122 is broken[19:45:39] <howarth> so this issue dates all the way back to at least 2013-02-11[19:50:08] <howarth> goes back to 2011-12-30 04:09:57[19:50:37] <howarth> so this has been broken for a really long time in master but masked by the reusage of the 1.5 branch for 1.5.1[19:50:46] <howarth> which was a really bone-headed move[19:51:33] *** jack has quit IRC[19:52:31] *** jack has joined #fink[19:52:55] *** jack is now known as Guest97120[19:55:00] <howarth> broken since 2011-06-29 23:38:10[19:59:25] *** Guest97120 is now known as jack[19:59:35] *** jack has quit IRC[19:59:35] *** jack has joined #fink[20:06:03] *** jack has quit IRC[20:08:52] <howarth> bingo[20:09:06] <howarth> looks like http://git.savannah.gnu.org/cgit/dejagnu.git/commit/?id=6d2e2d3791bcea70131a6cf64a0a5223333a7b8e is the offending commit[20:09:11] <howarth> which doesn't exist in 1.5.1[20:21:04] <howarth> confirmed that reverting that commit in 1.5.2 eliminates the bug[20:21:06] <howarth> nice[20:34:30] *** finkVC has joined #fink[20:34:30] <finkVC> jwhowarth: dists/10.7/stable/main/finkinfo/utils dejagnu.patch[20:34:30] <finkVC> add patch to revert commit that breaks gcc summaries and add upstream fixes for gcc[20:36:14] <finkVC> jwhowarth: dists/10.7/stable/main/finkinfo/utils dejagnu.info[20:36:16] <finkVC> add patch to revert commit that breaks gcc summaries and add upstream fixes for gcc[20:36:42] *** mischi has joined #fink[20:37:11] <finkVC> jwhowarth: dists/10.7/stable/main/finkinfo/utils dejagnu.info[20:37:13] <finkVC> add missing revision bump[21:30:33] *** asmeurer has joined #fink[21:50:19] *** mischi has quit IRC[22:00:27] *** akh has joined #fink[22:40:21] *** echelog-1 has joined #fink[22:51:08] *** asmeurer has joined #fink[22:57:17] *** asmeurer has quit IRC[23:13:16] *** asmeurer has joined #fink[23:23:23] *** asmeurer has quit IRC[23:27:20] *** mischi has joined #fink[23:37:22] *** finkVC has quit IRC