From: Gary K. <gw...@me...> - 2009-04-30 03:23:21
|
I've (finally) created a git repository of what was on SourceForge as of a few days ago. You can clone it using git clone http://common-lisp.net/project/asdf/asdf.git asdf The four of you should also be able to clone via git clone http://gk...@co.../project/asdf/asdf.git asdf Please let me know if anything goes awry. I plan to get the website up on cl.net soon (this weekend) and point the source forge pages to it. I'll include notes on the initial workflow too. Any comments on the setup, etc (or anything else) will be very welcome. thanks, -- Gary Warren King, metabang.com Cell: (413) 559 8738 Fax: (206) 338-4052 gwkkwg on Skype * garethsan on AIM * gwking on twitter |
From: Nikodemus S. <nik...@ra...> - 2009-04-30 05:42:52
|
2009/4/30 Gary King <gw...@me...>: > I've (finally) created a git repository of what was on SourceForge as of a > few days ago. You can clone it using > > git clone http://common-lisp.net/project/asdf/asdf.git asdf Or... git clone ssh://<username>@common-lisp.net/project/asdf/public_html/asdf.git which sets up things for push correctly from the get go -- except that asdf group doesn't seem to have write access to the repo yet. Cheers, -- Nikodemus |
From: Gary K. <gw...@me...> - 2009-04-30 11:09:06
|
> git clone ssh://<username>@common-lisp.net/project/asdf/public_html/ > asdf.git > > which sets up things for push correctly from the get go -- except that > asdf group doesn't seem to have write access to the repo yet. > Thanks, yes. I realized that I'd neglected this step just as I was drifting off to sleep. It'll be corrected later today. -- Gary Warren King, metabang.com Cell: (413) 559 8738 Fax: (206) 338-4052 gwkkwg on Skype * garethsan on AIM * gwking on twitter |
From: Robert G. <rpg...@si...> - 2009-05-04 19:16:35
|
As I understand things, we should, if we are working on a particular improvement, move ourselves onto a branch, make our changes, and then offer up the changes for some flavor of review, right? Can someone explain to me how the final step is done? I have only done this using org, where my changes were small enough that they could be eyeballed in patch files. But I have been working to get Marco's CLEAN-OP incorporated into ASDF, and this involves adding files (notably in order to make tests). This puts me, I think, beyond the realm of easily reviewed patches. In this case does one provide a branch in the common-lisp.net repo? Should one create one's own public repository? I have read some of the git tutorials but evidently I have not read the right ones, since the answers to these questions still elude me. I am very excited about the institution of the git repo, though, since the changes I am making, although they are modest (and most of the work is another's), they do involve non-local changes, and really seem to exceed what one can comfortably do with CVS. Best, R |
From: Nikodemus S. <nik...@ra...> - 2009-05-05 07:16:56
|
2009/5/4 Robert Goldman <rpg...@si...>: > As I understand things, we should, if we are working on a particular > improvement, move ourselves onto a branch, make our changes, and then > offer up the changes for some flavor of review, right? > > Can someone explain to me how the final step is done? I have only done > this using org, where my changes were small enough that they could be > eyeballed in patch files. Since we have not discussed it, I assume sending patches to the mailing list remains the preferred format. > But I have been working to get Marco's CLEAN-OP incorporated into ASDF, > and this involves adding files (notably in order to make tests). This > puts me, I think, beyond the realm of easily reviewed patches. Not really: with git those patches are best produced using "git format-patch -<small integer", which deals with new files just fine. > In this case does one provide a branch in the common-lisp.net repo? Should one > create one's own public repository? To the first, I would say no. I would suggest that we don't add feature branches to the main repository. Adding an "unstable" branch might be worth considering, though -- and put new features and API changes there initially. To the second, I would say that it's not really necessary. Not unless you feel that your work is being ignored, or unless you know that there are people who want to pull from yourself, or because you figure you want a backup for your local tree... Cheers, -- Nikodemus |
From: Gary K. <gw...@me...> - 2009-05-05 12:45:15
|
Hi Robert, > As I understand things, we should, if we are working on a particular > improvement, move ourselves onto a branch, make our changes, and then > offer up the changes for some flavor of review, right? I'm pretty new to git myself but am trying to use something like this workflow http://reinh.com/blog/2009/03/02/a-git-workflow-for-agile-teams.html > Can someone explain to me how the final step is done? I have only > done > this using org, where my changes were small enough that they could be > eyeballed in patch files. A Nikodemus suggested, I think sending a patch to the mailing list would be the best thing for now. (Once the repo is straightened out), I am going to add an unstable branch into which patches will go first. > But I have been working to get Marco's CLEAN-OP incorporated into > ASDF, > and this involves adding files (notably in order to make tests). This > puts me, I think, beyond the realm of easily reviewed patches. In > this > case does one provide a branch in the common-lisp.net repo? Should > one > create one's own public repository? I agree with Nikodemus (again! <smile>) > I have read some of the git tutorials but evidently I have not read > the > right ones, since the answers to these questions still elude me. I think that git is so flexible that knowing exactly what to do when is elusive. We'll have to create a path by walking. > I am very excited about the institution of the git repo, though, since > the changes I am making, although they are modest (and most of the > work > is another's), they do involve non-local changes, and really seem to > exceed what one can comfortably do with CVS. Yeah! Why don't you send the patch when you're ready and we can try to make asdf:clean-op a test case. -- Gary Warren King, metabang.com Cell: (413) 559 8738 Fax: (206) 338-4052 gwkkwg on Skype * garethsan on AIM * gwking on twitter |
From: Robert G. <rpg...@si...> - 2009-05-05 13:20:54
|
Gary King wrote: > Hi Robert, > >> As I understand things, we should, if we are working on a particular >> improvement, move ourselves onto a branch, make our changes, and then >> offer up the changes for some flavor of review, right? > > I'm pretty new to git myself but am trying to use something like this > workflow > > http://reinh.com/blog/2009/03/02/a-git-workflow-for-agile-teams.html > >> Can someone explain to me how the final step is done? I have only >> done >> this using org, where my changes were small enough that they could be >> eyeballed in patch files. > > A Nikodemus suggested, I think sending a patch to the mailing list > would be the best thing for now. My concern arose about shipping around patches involving the addition of multiple files (this arises when creating new tests for asdf). [...snip...] > >> I am very excited about the institution of the git repo, though, since >> the changes I am making, although they are modest (and most of the >> work >> is another's), they do involve non-local changes, and really seem to >> exceed what one can comfortably do with CVS. > > Yeah! Why don't you send the patch when you're ready and we can try to > make asdf:clean-op a test case. Will do. I shelved this for a long while because I wasn't sure how to deal with creating the new test cases with only CVS to rely on. Might take me a little while to recover the context. Best, R |
From: Nikodemus S. <nik...@ra...> - 2009-05-05 07:28:11
|
The repo still doesn't seem set up quite right. When I try to push, I get: Counting objects: 5, done. Compressing objects: 100% (3/3), done. Writing objects: 100% (3/3), 382 bytes, done. Total 3 (delta 2), reused 0 (delta 0) error: Unable to append to logs/HEAD: Permission denied To ssh://nsi...@co.../project/asdf/public_html/asdf.git c457de7..ddcf33a master -> master I believe this is because core.sharedRepository is not set correctly. See "git help config": http://www.kernel.org/pub/software/scm/git/docs/git-config.html Cheers, -- Nikodemus |
From: Nikodemus S. <nik...@ra...> - 2009-05-05 08:45:52
|
2009/5/5 Nikodemus Siivola <nik...@ra...>: > The repo still doesn't seem set up quite right. When I try to push, I get: > > Counting objects: 5, done. > Compressing objects: 100% (3/3), done. > Writing objects: 100% (3/3), 382 bytes, done. > Total 3 (delta 2), reused 0 (delta 0) > error: Unable to append to logs/HEAD: Permission denied > To ssh://nsi...@co.../project/asdf/public_html/asdf.git > c457de7..ddcf33a master -> master > > I believe this is because core.sharedRepository is not set correctly. > See "git help config": > > http://www.kernel.org/pub/software/scm/git/docs/git-config.html Digging deeper, I'm guessing you didn't initialize the repo using --shared=group at all, so the permissions didn't get set right in the first place, and I think setting that flag (which I managed to do) is not enough to fix them. I made a copy in ~nsiivola on common-lisp.net, and tried to fix the permissions there, and while I think I got them right, I cannot replace the main repository, nor can I fix its permissions as the project directory itself is missing group write permissions, and the public_html directory is missing group write and sticky bit. As for asdf.git, I believe the correct permissions are as follows: For the directory and all subdirectories: dwrxrwsr-x (note the group sticky bit) For files under objects/??/ and objects/pack/: -r--r--r-- For files under hooks/, -rwxrwxr-x For all other files, including those under objects/info/: -rw-rw-r-- ...and the group should be asdf everywhere, of course (it's currently users on a few subdirectories.) (I managed to fix the complete one push by accident, since I was allowed to create *new* directories under objects/, and was able to fix some of the permissions. Other than the messed up permissions and ownerships, it looks healthy to me.) Cheers, -- Nikodemus |
From: Gary K. <gw...@me...> - 2009-05-05 13:02:38
|
> Digging deeper, I'm guessing you didn't initialize the repo using > --shared=group at all, so the permissions didn't get set right in the > first place, and I think setting that flag (which I managed to do) is > not enough to fix them. Whoopts. That's right. I didn't > I made a copy in ~nsiivola on common-lisp.net, and tried to fix the > permissions there, and while I think I got them right, I cannot > replace the main repository, nor can I fix its permissions as the > project directory itself is missing group write permissions, and the > public_html directory is missing group write and sticky bit. > Fixed (I think) > As for asdf.git, I believe the correct permissions are as follows: > > For the directory and all subdirectories: dwrxrwsr-x (note the group > sticky bit) ok > > For files under objects/??/ and objects/pack/: -r--r--r-- ok > > For files under hooks/, -rwxrwxr-x ok > > For all other files, including those under objects/info/: -rw-rw-r-- ok > > > ...and the group should be asdf everywhere, of course (it's currently > users on a few subdirectories.) > ok > (I managed to fix the complete one push by accident, since I was > allowed to create *new* directories under objects/, and was able to > fix some of the permissions. Other than the messed up permissions and > ownerships, it looks healthy to me.) > Can you please check again. What about these three new files > -rw-rwSr-- 1 nsiivola asdf 117 2009-05-05 03:44 config > -rw-rwSr-- 1 nsiivola asdf 58 2009-05-05 03:44 description > -rw-rwSr-- 1 nsiivola asdf 23 2009-05-05 03:44 HEAD Let me know if I should just try creating it again... -- Gary Warren King, metabang.com Cell: (413) 559 8738 Fax: (206) 338-4052 gwkkwg on Skype * garethsan on AIM * gwking on twitter |
From: Nikodemus S. <nik...@ra...> - 2009-05-05 16:09:50
|
2009/5/5 Gary King <gw...@me...>: >> replace the main repository, nor can I fix its permissions as the >> project directory itself is missing group write permissions, and the >> public_html directory is missing group write and sticky bit. > Fixed (I think) No group write in the project directory: drwxr-sr-x 4 mbaringer asdf 4096 2009-05-05 08:48 /project/asdf/ should be drwxrwsr-x. No group sticky bit in public_html (instead deletion is restricted to owner): drwxrwxr-t 4 gking asdf 4096 2009-05-05 11:03 /project/asdf/public_html/ should be drwxrwsr-x as well. >> For the directory and all subdirectories: dwrxrwsr-x (note the group >> sticky bit) > > ok Enough of the asdf.git permissions are still out of whack to render the repo inoperable (missing write and directory access permissions, plus unhygienic write permissions in places where there should be none.) A fixed repo, which should have the group permissions right is now at /project/asdf/public_html/asdf-new.git -- it has the same objects as the earlier one, plus one new commit on top. I would have replaced the broken one, but due to the t bit in public_html/ I still could not. If you can verify that this one works for you (test pushing as well...), then just renaming it to asdf.git should have us up and running. For future reference, I belatedly realized that the easiest way to fix things was to make a new empty repo that had the right group and was initialized with --shared=group, and then push everything from a working clone there: cd ~/ newgrp asdf mkdir asdf.git chmod g+ws asdf.git cd asdf.git git init --bare --shared=group and then from my home machine: cd src/asdf-git # a working clone made from the old asdf.git git push ssh://nsi...@co.../home/nsiivola/asdf.git master and finally on common-lisp.net cd /project/asdf/public_html mv ~/asdf.git asdf-new.git Cheers, -- Nikodemus |
From: Gary K. <gw...@me...> - 2009-05-05 17:52:31
|
Got 'em. (I think I was (am!) confused about sticky bit versus setuid bit. We want the latter, not the former, right? On May 5, 2009, at 12:09 PM, Nikodemus Siivola wrote: > 2009/5/5 Gary King <gw...@me...>: > >>> replace the main repository, nor can I fix its permissions as the >>> project directory itself is missing group write permissions, and the >>> public_html directory is missing group write and sticky bit. > >> Fixed (I think) > > No group write in the project directory: > > drwxr-sr-x 4 mbaringer asdf 4096 2009-05-05 08:48 /project/asdf/ > > should be drwxrwsr-x. > > No group sticky bit in public_html (instead deletion is restricted > to owner): > > drwxrwxr-t 4 gking asdf 4096 2009-05-05 11:03 /project/asdf/ > public_html/ > > should be drwxrwsr-x as well. > >>> For the directory and all subdirectories: dwrxrwsr-x (note the group >>> sticky bit) >> >> ok > > Enough of the asdf.git permissions are still out of whack to render > the repo inoperable (missing write and directory access permissions, > plus unhygienic write permissions in places where there should be > none.) > > A fixed repo, which should have the group permissions right is now at > > /project/asdf/public_html/asdf-new.git > > -- it has the same objects as the earlier one, plus one new commit on > top. I would have replaced the broken one, but due to the t bit in > public_html/ I still could not. If you can verify that this one works > for you (test pushing as well...), then just renaming it to asdf.git > should have us up and running. > > For future reference, I belatedly realized that the easiest way to fix > things was to make a new empty repo that had the right group and was > initialized with --shared=group, and then push everything from a > working clone there: > > cd ~/ > newgrp asdf > mkdir asdf.git > chmod g+ws asdf.git > cd asdf.git > git init --bare --shared=group > > and then from my home machine: > > cd src/asdf-git # a working clone made from the old asdf.git > git push ssh://nsi...@co.../home/nsiivola/asdf.git master > > and finally on common-lisp.net > > cd /project/asdf/public_html > mv ~/asdf.git asdf-new.git > > Cheers, > > -- Nikodemus > > ------------------------------------------------------------------------------ > The NEW KODAK i700 Series Scanners deliver under ANY circumstances! > Your > production scanning environment may not be a perfect world - but > thanks to > Kodak, there's a perfect scanner to get the job done! With the NEW > KODAK i700 > Series Scanner you'll get full speed at 300 dpi even with all image > processing features enabled. http://p.sf.net/sfu/kodak-com > _______________________________________________ > cclan-list mailing list > ccl...@li... > https://lists.sourceforge.net/lists/listinfo/cclan-list -- Gary Warren King, metabang.com Cell: (413) 559 8738 Fax: (206) 338-4052 gwkkwg on Skype * garethsan on AIM * gwking on twitter |
From: Nikodemus S. <nik...@ra...> - 2009-05-05 18:39:41
|
2009/5/5 Gary King <gw...@me...>: > Got 'em. > > (I think I was (am!) confused about sticky bit versus setuid bit. We want > the latter, not the former, right? Oops, actually, my terms were reversed. We want the SGID, aka set group id bit on directories -- not the sticky bit. It's right in the asdf-new.git tree, it's the bit that shows up as an 's' in the group permissions when combined with exec permissions. Those bits are as they are supposed to be in the asdf-new.git tree. (We want it in order to have files created in those directories have the same group as the directory, as opposed to whatever happens to be the primary group of the user at that moment.) Cheers, -- Nikodemus |