| <<<Back 1 day (to 2012/11/17) | 2012/11/18 |
ray_laptop | Robin_Watts: you probably are not working, but ... | 04:58.17 |
| Robin_Watts: in case you are. | 04:58.26 |
| I have a local git repo for my customer 532 code (which is 8.71+++) and I wanted to 'stash' some WIP. but when I did " git stash save "limit_image_interp" I got: | 05:01.19 |
| fatal: ambiguous argument 'HEAD': both revision and filename | 05:01.20 |
| Use '--' to separate filenames from revisions | 05:01.21 |
| Saved working directory and index state On (no branch): limit_image_interp | 05:01.23 |
| HEAD is now at 715d318 Force interpolation to the parser stage | 05:01.24 |
| now my WIP is gone ! (except from my physical head). | 05:01.56 |
| so, other than resorting to "wet memory" are my changes laying around somewhere ? | 05:02.37 |
| note that: git stash list shows nothing and "git status -u no" shows: | 05:05.11 |
| # Not currently on any branch. | 05:05.13 |
| nothing to commit (working directory clean) | 05:05.14 |
| prior to the git stash save, "git st" showed: | 05:06.29 |
| M base/gxclimag.c | 05:06.31 |
| M base/gxi12bit.c | 05:06.32 |
| M base/gxiclass.h | 05:06.34 |
| M base/gxicolor.c | 05:06.35 |
| M base/gxidata.c | 05:06.37 |
| M base/gxifast.c | 05:06.39 |
| M base/gximage.h | 05:06.40 |
| M base/gximono.c | 05:06.42 |
| M base/gxipixel.c | 05:06.43 |
| M base/gxiscale.c | 05:06.45 |
| M base/icclib.mak | 05:06.46 |
| M base/jpeg.mak | 05:06.48 |
| M base/ktd_msvclib.mak | 05:06.49 |
| M base/winlib.mak | 05:06.51 |
| M psi/int.mak | 05:06.52 |
| M psi/ktd_gs_main.c | 05:06.54 |
| M psi/msvc.mak | 05:06.55 |
sebras | rayjj: if you have file .git/refs/stash with a commit id in it, try to checkout out that commit id. this is apparently where git normally stores the sha1 of things you stash. | 14:13.56 |
| rayjj: though I would expect that "branch" to show up in gitk/git logg/whatever and in that case you would probably have found it already. | 14:14.42 |
| rayjj: I tried to reproduce your problem (I guessed you had a branch named limit_image_interp as well as naming the stash this), but using git 1.7.10.4 I cannot reproduce your error message. | 14:15.44 |
ray_laptop | my local .git LOST my changes. I search "objects" and I do see a couple of fairly recent files that look like SHA1 names. How do I load them to see what's in them ? | 19:16.41 |
| This is a result of doing a stash, but my stash is gone (there isn't even a .git/refs/stash directory) | 19:17.48 |
saper | ray_laptop: you can read the object file with "zpipe", a simple C program that calls deflate() from zlib | 19:18.41 |
| ray_laptop: the contents is fairy simple, but you probably only need to do "git show <...sha1...>" to get the object | 19:19.29 |
| ray_laptop: what do you mean by "LOST" my changes? maybe you were switched to another branch and your commits just are not plain visible to "git log"? | 19:20.02 |
| to check that, I would recommend having a look at "git reflog" - it shows history of movements of your current head | 19:20.35 |
ray_laptop | saper: thanks. trying git show first ... | 19:22.09 |
| the most recent SHA1 files don't show up on git reflog at all | 19:23.17 |
saper | sha1's objects in the objects/ are also trees and blobs, which you won't see | 19:28.47 |
| (only objects of type "commit" are) | 19:29.05 |
ray_laptop | I had 17 files that "git st" showed as modified. I did 'git stash save "limit_image_interp"'. It said: | 19:31.57 |
| fatal: ambiguous argument 'HEAD': both revision and filename | 19:31.59 |
| Use '--' to separate filenames from revisions | 19:32.00 |
| Saved working directory and index state On (no branch): limit_image_interp | 19:32.02 |
| HEAD is now at 715d318 Force interpolation to the parser stage | 19:32.03 |
| I didn't understand the 'fatal' message, but since it said it 'saved' it, I didn't worry overmuch. | 19:32.49 |
| so is there a way to get back what it said it saved ??? | 19:34.08 |
sebras | ray_laptop: does git reflog show you any sha1 that contains the files you are looking for? | 19:41.45 |
| ray_laptop: it might, but I think that git stash sha1s do not show up there. :-/ | 19:42.07 |
ray_laptop | sebras: nope. git reflog just shows: | 19:43.22 |
| 715d318 HEAD@{0}: commit: Force interpolation to the parser stage | 19:43.24 |
| 91976f1 HEAD@{1}: commit: Increase MAX_BITMAP_PATTERN_SIZE to 6m and force pattern-clists to non HL images. | 19:43.25 |
| fa7fb75 HEAD@{2}: checkout: moving from master to HEAD~1 | 19:43.27 |
| 1ca498a HEAD@{3}: commit: Port the 9.06 changes to support overprint to RGB devices. | 19:43.28 |
| fa7fb75 HEAD@{4}: commit: Fix possible access past array bounds. | 19:43.30 |
sebras | ray_laptop: right, and those commit ids are just normal commits as I understand..? | 19:45.28 |
| ray_laptop: have you continued to work in that repo after the problem happened? | 19:45.42 |
ray_laptop | sebras: no | 19:45.52 |
sebras | ray_laptop: is not one could possibly try to dig out the sha1s from .git/objects/* based on time perhaps. | 19:46.02 |
| the directories contain the two first hexdigits of the sha1 and the filenames the remainder. | 19:46.32 |
ray_laptop | there are only a few files with Nov 17 date (all the rest are older) | 19:47.03 |
sebras | ray_laptop: I need to go to get some food, but this might be a good resource http://ftp.newartisans.com/pub/git.from.bottom.up.pdf | 19:47.48 |
ray_laptop | sebras: AHH.. so I need to concatenate the first 2 and the remainder in order to actually form the SHA1 | 19:48.01 |
sebras | ray_laptop: exacrtly. | 19:48.40 |
| thought the sha1 might be one for a particular file or for a tree, not necessarily a commit id. | 19:49.00 |
| look in the pdf and the datastructures will make more sense. :) | 19:49.19 |
sebras | rushes starving to the local pizza place... ;) | 19:49.42 |
saper | this wonders me: " On (no branch)" | 19:55.33 |
| were you in the so-called detached HEAD? (it happens if you "git checkout <sha1>", nothing too much to worry about) | 19:56.03 |
sebras | ray_laptop: do you have a branch or a file named HEAD in that repo? | 22:15.59 |
| ray_laptop: judging by the error messages that's what it looks like. | 22:16.24 |
itwasntme | ghostscript: the first scripting language for ghosts. where do i begin? | 23:34.01 |
| oh...i see | 23:34.24 |
| Forward 1 day (to 2012/11/19)>>> | |