IRC log for #koha, 2007-11-13

All times shown according to UTC.

Time S Nick Message
17:49 slef fbcit: run git update-server-info and then rsync your .git to somewhere in your web space every so often... that's the simplest and dumbest way to do it, but uses a lot of webspace
17:49 slef s/rsync/mirror
17:50 slef got to go... back in a bit
17:51 kados so record.abs is so-named, at least in theory, because it's used to parse <record>s
17:51 kados I wonder if the latest version of zebra allows you to change the name
17:51 kados if so we could have both in the same dir
17:51 fbcit slef: I'll try to get my local tree published...
17:53 fbcit kados: I got my clone screwed up again and it will not rebase.... >8-(
17:53 atz fbcit: I seem to be good at that too
17:55 fbcit atz: I do an install off of my clone; mod the install; test; apply to clone; commit... is that sane?
17:56 atz i tend to just use 1 layer... i.e. mod the clone
17:57 atz you can get hierarchical w/ git branches and everything... but that's overkill for me right now
17:58 fbcit about every other day I git a local patch that will not apply when doing rebase.
17:58 kados fbcit: you can run your install directly off of git
17:58 kados fbcit: that's what we do internally for testing
17:59 fbcit so... mod clone; install off of clone; test?
17:59 kados it's more like
17:59 kados mod clone; test
17:59 kados since you're running directly off the clone
17:59 kados all you need to do is set up a few symlinks
17:59 fbcit ahh...
18:01 kados basically link the cgi-bin dirs directly to the /koha clone
18:01 kados and the htdocs to /koha/koha-tmpl
18:01 kados and modules/C4 to koha/C4
18:02 kados you can even get fancy and link some of the config files too
18:03 kados where it created those symlinks for you from a git clone
18:03 paul superb idea kados !!!
18:04 kados 'Koha Developer's Toolkit'
18:04 fbcit is that a Feature Request? :-)
18:05 fbcit I think the installer needs to offer to install/update deps...
18:21 kados hdl: you around?
18:21 kados hdl: if I have independantbranches set to ON, should I be able to 'Set' my branch?
18:29 fbcit kados: It appears that I have a conflict with a file that changed in HEAD and was patched as deleted in my clone.
18:29 fbcit Is there a way to force git to apply my patch locally anyway?
18:29 kados hmmm
18:30 kados I'm not sure
18:30 kados walk through what you did?
18:31 fbcit I moved a file (several actually) to a different location...
18:31 fbcit I committed the old file....
18:31 fbcit added the new file....
18:31 fbcit patched...
18:31 fbcit rebased...
18:32 fbcit and ...
18:32 fbcit CONFLICT (delete/modify): installer/data/en/mandatory/sysprefs.sql deleted in Moved for multi-dbms scheme and modified in HEAD. Version HEAD of installer/data/en/mandatory/sysprefs.sql left in tree.
18:32 kados did you use git-mv
18:32 fbcit I had to move the data/en, data/fr, to data/'dbms'/en, etc...
18:33 kados and git-rm?
18:33 fbcit no... git commit /path/to/old/file
18:33 fbcit git add /path/to/new/file
18:33 kados ok, lets try something
18:34 kados when you make your patch
18:34 kados go:
18:34 kados git-format-patch -M -B origin
18:34 kados that will format it special for handling moves and deletes
18:35 fbcit is it possible to re-make patches? I did git format-patch origin the other night/morning....
18:35 kados yea
18:35 fbcit k
18:35 kados just delete or move the patch files out of the way
18:35 kados and run it again
18:36 fbcit that reduced my patch count from 26 to 6...
18:36 fbcit now rebase?
18:38 kados hmmm
18:38 kados rebasing should have reduced the patch count
18:38 kados if some of your patches were accepted
18:38 kados but I don't think we've seen 20 patches from you
18:39 fbcit part of them may have been, but it hung on #7 of 26 I think...
18:39 fbcit no..
18:40 fbcit hdl moved some of the same files and some of those moves matched mine I think... his patches were accepted....
18:40 fbcit shall I try rebase?
18:40 kados ahh, that would explain it then
18:40 kados what I'd do is:
18:40 kados git status
18:40 kados that will tell you if you have changes that haven't been committed
18:40 kados if there are none
18:40 kados git fetch
18:41 kados git rebase
18:41 kados git rebase origin I mean
18:41 kados that should merge all the upstream changes and reapply your patches
18:41 kados to the clone/branch you've got checked out
18:41 kados then ...
18:41 kados you want to make your patches
18:41 kados git-format-patch -M -B origi
18:41 kados n
18:41 fbcit I have some "changed but not updated"
18:42 fbcit hrmm...
18:42 kados ok, that means you have modifcations you haven't committed
18:42 kados so first step is to commit those
18:43 kados git commit -a
18:43 kados will catch all fo them
18:43 kados all of them even
18:45 fbcit k
18:45 fbcit had to do git add
18:46 fbcit git commit said 'nothing added to commit...'
18:46 fbcit now git status has no changes listed... only untracked files...
18:46 fbcit so git fetch?
18:46 kados ok, cool
18:46 kados so git commit -a
18:46 kados doesn't do anything?
18:47 fbcit no. It appears that a file needed to be added but had no changes... ???
18:47 fbcit now git rebase origin...
18:48 kados git fetch first
18:48 kados that will update your repo
18:48 fbcit right. I did...but
18:48 fbcit no go.
18:48 kados got rebase origin
18:48 kados git rebase origin I mean
18:48 fbcit It still does not that patch.
18:49 kados so what's the error?
18:49 fbcit installer/data/en/mandatory/sysprefs.sql deleted in Moved for multi-dbms scheme and modified in HEAD. Version HEAD of installer/data/en/mandatory/sysprefs.sql left in tree.
18:50 kados ok, so you can do
18:50 kados git whatchanged
18:50 fbcit Which is not even in the lsit of patches now
18:50 kados and find that commit
18:50 fbcit s/lsit/list
18:50 kados well, if your rebase origin failed
18:51 kados you haven't re-applied all of the commits
18:51 kados hmmm
18:51 kados we might need to consult the git gods :-)
18:51 kados freenode #git
18:53 kados fbcit: I'm asking on freenode
18:53 kados fbcit: any more to that error message?
18:54 fbcit that's it...
18:55 fbcit actually it is prefixed with: CONFLICT (delete/modify):
18:56 kados hmmm
18:57 kados so it just stops then?
18:57 kados what happens if you run it again?
18:57 fbcit Failed to merge in the changes.
18:57 fbcit Patch failed at 0007.
18:57 fbcit When you have resolved this problem run "git rebase --continue".
18:57 fbcit If you would prefer to skip this patch, instead run "git rebase --skip".
18:57 fbcit To restore the original branch and stop rebasing run "git rebase --abort".
18:58 fbcit I just tried --skip
18:58 kados yea, that seems good
18:58 kados you can always re-do those changes
18:58 fbcit Dirty index: cannot apply patches (dirty: installer/data/en/mandatory/sysprefs.sql)
18:58 kados hmmm
18:58 fbcit :-(
18:58 kados that's what you get when you do git rebase --skip
18:58 kados ?
18:58 fbcit yes
18:59 kados ok, so what was the anture of the change wtih installer/data/en/mandatory/sysprefs.sql?
18:59 kados nature
19:00 kados hmmm
19:00 fbcit It appears that I have deleted it (moved it) and someone has modified it in the HEAD
19:00 kados right
19:00 fbcit so git cannot see it needs to apply the change from HEAD and then move because I did a commit deleted file, add new file...
19:01 kados yea
19:01 kados git-mv would have been better I think
19:01 fbcit that probably would allow git to track the move....
19:02 kados yea
19:02 kados try this:
19:02 fbcit I did this for all of the files under installer/language... so things are really screwed up...
19:02 kados git add installer/data/en/mandatory/sysprefs.sq
19:02 kados l
19:02 kados git rebase --continue
19:03 kados ahh, hang on
19:03 kados kados: easiest thing would be to go back where you started the rebase, and try with --merge.
19:03 kados from #git
19:04 kados so I _think_ if you do a 'git checkout branchname'
19:04 kados that will put you back to where you were
19:04 kados and we can try rebasing with the --merge option
19:05 fbcit so... git checkout branch?
19:05 kados hmmm, no that won't work
19:05 kados shoot
19:05 fbcit --merge did not work either... same error...
19:06 kados maybe slef knows what to do
19:06 kados slef: fbcit is in git hell
19:06 slef "Server load is temporarily too heavy. Please wait a while and try again.
19:06 slef Disconnected
19:06 slef "
19:06 slef never seen that before
19:06 kados slef: he is trying to rebase and it's failing, someone made a change in a file that he deleted
19:06 fbcit what if I do git-mv on the file in question and then rebase?
19:07 slef kados: hrm... delete it again, git update-index then git rebase --continue?
19:07 slef fbcit: do you want the file or not?
19:08 fbcit no...
19:08 kados I think it should be moved to a new location
19:08 kados unless it's in both places now
19:08 fbcit hold on..
19:09 fbcit no it is in another place.
19:09 fbcit when I git rebase --abort it appears where I moved it to...
19:10 fbcit when I git rebase origin, it shows up in the same location as HEAD....
19:10 fbcit maybe just git update-index as slef suggested?
19:11 slef check git status
19:11 slef see if it looks close to what you want
19:11 fbcit looks good
19:14 slef chris: sorry!
19:14 fbcit git update-index now?
19:14 slef fbcit: yep, if you want to confirm the changes shown in git status
19:14 slef git update-index whatever/file/name
19:17 fbcit no go... still errors on the same patch.
19:19 slef erm, haven't you already applied that patch and resolved the conflict?
19:19 slef maybe I'm not clear on how you're rebase-ing
19:20 fbcit git rebase origin
19:20 fbcit after git format-patch
19:23 kados fbcit: did you try formatting your patches with those extra flags?
19:23 fbcit yes
19:24 fbcit kados: btw, I have all 27 patches again... it appears I did the prevous format-patch before aborting the rebase.
19:24 slef I don't understand, sorry.
19:25 slef What has format-patch got to do with rebase?
19:25 fbcit slef: it appears that when a rebase "hangs" one must either --continue, --skip, or --abort...
19:26 fbcit until one of those choices is executed the branch is current with the last successfully applied patch...
19:26 fbcit so when I format-patch'd before rebase --abort, only 7 patches formatted up...
19:26 slef you commit on your topic-branch, checkout master, pull, rebase master topic-branch... if rebase hits a conflict, then fix it and rerun the rebase with --continue
19:27 fbcit right... how to fix this conflict is the question...
19:27 slef another option for more complex work is to keep a local branch and then merge master into it
19:28 slef which I think is what I should start doing
19:31 fbcit slef: rebase still runs into a conflict with this file...
19:32 slef what is error?
19:32 fbcit CONFLICT (delete/modify): installer/data/en/mandatory/sysprefs.sql deleted in Moved for multi-dbms scheme and modified in HEAD. Version HEAD of installer/data/en/mandatory/sysprefs.sql left in tree.
19:36 slef hrm, seems like something isn't detecting a mv...
19:36 slef what's git's advice?
19:36 slef as in, what does it suggest you should do?
19:36 fbcit When you have resolved this problem run "git rebase --continue".
19:37 fbcit If you would prefer to skip this patch, instead run "git rebase --skip".
19:37 fbcit To restore the original branch and stop rebasing run "git rebase --abort".
19:37 slef Personally, I'd git rm it and rebase --continue master branch ; failing that --skip and try to sort it out later
19:41 fbcit hold on.... #git is speaking...
19:41 slef and it says "phrrrrp"?
19:43 fbcit :-)
19:48 fbcit :q
19:50 slef so what do the sages say?
19:51 fbcit their suggestion was to manually apply the changes...
19:53 fbcit it appears to me that git wants to apply my patch (remove the file) and then apply the HEAD diff which mod's the file. But that seems backwards....
19:53 fbcit #git's suggestion did not work....
19:53 slef :-(
19:56 fbcit when I do git checkout -b foo
19:56 fbcit is foo based on master or origin?
19:57 slef it's based on whatever you had checked out I think
19:57 slef before that, I mean
19:58 gmcharlt it's whatever your current HEAD is, according to doc
19:59 slef I'm probably going to regret asking, but: current HEAD?
19:59 slef as in *current*?
19:59 fbcit I guess I'll just blow away the entire thing and start over. Now when I checkout a new branch, it has the same issues as the existing one... :-P
20:00 gmcharlt slef: current in your local repo, I assume
20:10 fbcit ok what is the "correct" way to move files from one dir to another in git?
20:10 slef git mv?
20:11 fbcit slef: tnx for the help.
20:11 slef ,:-/
20:11 fbcit hdl: ping
20:12 slef I think the frenchies have gone home
20:52 fbcit kados: ping
21:09 owen Any git experts awake?
21:15 kados hiya fbcit, owen
21:16 owen Hi kados, I was just trying to figure out a git conflict message
21:16 kados sure
21:17 kados so what's the error you're getting?
21:17 kados heeh
21:17 owen Actually, it's saying something about a file being renamed in HEAD and removed in one of my commits
21:18 kados hehe
21:18 kados fbcit just went through this I think
21:18 kados fbcit: what was the final solution?
21:19 fbcit fetch a new clone of into kohaclone1 and start moving my changes over to the new clone... :-(
21:19 kados ouch
21:20 kados owen: what file is it?
21:20 fbcit I think the removing of a file locally which is then modified at HEAD prior to a rebase puts git into a tight-loop...
21:20 owen Here's the message: "Renamed koha-tmpl/intranet-tmpl/prog/en/includes/boo​>koha-tmpl/intranet-​tmpl/prog/en/includes/ in HEAD and deleted in Lots of work to standardize elements... <snip>"
21:21 kados huh
21:21 kados I don't remember those being deleted
21:22 kados owen: can you paste in the whole message?
21:24 owen The whole rebase message?
21:24 kados just from where it started complaining
21:24 fbcit kados: can 'git format-patch -M -B origin' be used for all patches?
21:25 kados fbcit: dont think it's good for all patches
21:25 kados IIRC, it causes problems if you didn't move or delete files
21:25 fbcit k so only on those doing moves...
21:25 fbcit wonder how I distinguish between them?
21:26 kados yea, I dunno
21:26 fbcit I have some that are moves and some that are code changes...
21:26 kados question for #git I suppose
21:26 kados *nod*
21:26 kados might have to push them up in iterations
21:26 fbcit the learning curve is killing me.
21:26 kados yea, us too
21:26 kados when it works well it rocks
21:26 kados when it doesn't work well it really bites
21:27 fbcit I have 8 patches against my new clone
21:27 kados can eat up al lyour time
21:27 fbcit they are a mixture...
21:27 kados send em through with the flags
21:27 kados and see what you get back from chris
21:27 fbcit k
21:28 owen kados, I just emailed you the whole message, since I'm not really sure what's relevant
21:28 kados ok, I'll have a look
21:29 kados hmmm
21:29 kados owen: so you deleted some files, eh?
21:29 owen Yes
21:30 kados I wonder
21:30 kados what if you did a fresh clone
21:30 kados did git-rm on the same files
21:30 kados oops
21:30 kados yea, that's right
21:30 fbcit kados: is there a git command for mkdir?
21:30 kados fresch clone
21:30 kados git-rm on the same files
21:30 kados fbcit: no, git doesn't care about dirs, just files
21:31 kados owen: then make a patch for those rms and send them upstream
21:33 fbcit kados: I'm going send the patches up for the port so far. I think I have remembered to re-make all of my changes....
21:34 fbcit It works over mysql on my test box with the exception of a problem with one sql script which I think is a result of some of hdl's changes...
21:34 owen So kados, you're suggesting I figure out which files I removed, and send a patch containing just those removals?
21:35 fbcit I could wait until tomorrow and test again, but I am afraid to wait in light of this being the third time I have had to redo things....
21:35 fbcit If things break they should not break too bad I think.
21:36 fbcit I still have to report the marc framework scripts for pg so that portion will not work yet...
21:38 fbcit I don't have README.txt mod'd yet.
21:39 fbcit I'll tidy up tomorrow and let you know when the installer is ready to test over pg.
21:40 fbcit I think having the file location changes committed will help avoid the issue I ran into today.
21:41 fbcit g'night.
21:43 kados owen: yea, hopefully that will fix the merging
21:50 owen kados, the files I deleted in those patches are not in the fresh clone
22:15 kados arrg, sorry
22:15 kados phone call
22:15 kados damn customers
22:15 kados :-)
22:20 kados owen: any progress on your git probs?
22:28 owen No, I'll have to pick it up tomorrow
22:29 kados sorry :/
23:52 fbcit-1 kados: u around?
23:52 kados fbcit-1: yuppers
23:53 fbcit-1 I knew I would forget something...
23:53 kados fbcit-1: got your patches, I'll be testing them tomorrow morning
23:53 fbcit-1 I failed to submit a patch for the step2 template
23:53 kados ahh, sweet
23:53 fbcit-1 I'll try to do it first thing tomorrow...
23:53 kados send it to patches@
23:53 kados ok, no worries
23:53 kados I'll wait for that one to start testing
23:54 fbcit-1 k
23:54 fbcit-1 my brain was really frazzled this afternoon. :-(
23:55 fbcit-1 I was just reviewing the patches in my email and realized I forgot that one.
23:55 kados no worries
23:55 fbcit-1 kados: I also have to bring the pg version of kohastructure.sql up to date. I noticed several schema changes in the mysql version.
00:05 fbcit-1 kados: looks like the patch for step3.tmpl will need adjusting... it suffers from my copy being out of sync... sorry.
00:05 kados no worries, we'll get it sorted
00:05 fbcit-1 actually ,you need to disgard the entire patch. I'll have to resubmit it.
00:35 kados fbcit-1: resubmit all of them or just one/some?
00:36 fbcit-1 kados: just the one for step3.tmpl...
00:37 fbcit-1 as it turns out, that patch is only a diff between my version and the current with none of my changes to step3.tmpl in it.
00:38 fbcit-1 so it will only roll step3.tmpl backward if applied. I probably typo'd step3 and meant step2.... sorry.
00:40 fbcit-1 kados: I may be back in my office late tonight.
00:40 fbcit-1 If so, I'll submit the correct patch.
00:40 kados sweet
00:40 kados fbcit-1++ # you rock
01:48 fbcit kados:ping
01:48 fbcit I submitted a patch to reverse the patch for step3.html I submitted earlier.
01:49 fbcit s/html/tmpl/
01:49 fbcit I also submitted the correct patch for step2.tmpl
01:50 fbcit you should be good to test tomorrow morning for breaks to the mysql install process.
02:00 kados fbcit: sweet, thanks
03:08 fbcit maybe 'git' will cooperate this time....
03:11 kados w00t
03:27 fbcit with only two complaints about duplicate entries while importing sql...
03:28 fbcit kados: the only thing left is the option sql under the english and then all of the sql under the french...
03:31 fbcit kados: something is strange about the issues with zebra lately. I run it just fine...
03:32 fbcit I looked at mourik's xml the other day and compared it to mine.... it was the same with expected exceptions...
03:35 fbcit git is messed up again...
03:35 fbcit fbcit:/home/fbcdev-1/kohaclone1# git rebase origin
03:35 fbcit cannot rebase: your index is not up-to-date
03:35 fbcit D       installer/kohastructure_pg.sql
03:36 fbcit so I'm going for the night.

Channels | #koha index | Today | | Search | Google Search | Plain-Text | plain, newest first | summary