Re: git hang-ups
Patrick McCarty wrote Saturday, August 08, 2009 5:00 AM On Fri, Aug 07, 2009 at 08:39:16PM -0700, Mark Polesky wrote: Patrick McCarty wrote: > I've just tested git's shallow cloning feature. It's pretty > neat. > :-) > > From what I can see, shallow clones would be okay for *casual* > contributors that are only sending patches based on the tip of > master. > > However, since git history is limited to the depth of the > clone, then > shallow clones would not permit a developer to revert a commit > from, > say, three weeks ago. > > In other words, I think both the "git clone" and "git > clone --depth" > methods should be included in the CG. To me, it seems that a developer should be able to just stick to shallow clones for everyday use. I assume that one could simply increase the depth of the clone when needed. Is that true? I've never needed to revert a three-week old commit, but if I needed to, I figure I would just do another clone with a greater depth. Does it work that way? I suppose that sounds reasonable, but if you "reclone" a repository two or three times, each time with greater depth, then you'll probably be using more bandwidth than if you had just downloaded the entire repo initially (with "git clone"). Really, it all depends on how developers use git history. Personally, I browse git history on the command line quite often when working with LilyPond. A while back, I needed to find a change made to a certain file several years ago (I don't remember which file). To exacerbate the problem, the file had been renamed once or twice. But git makes this easy: $ git log -p --follow file.scm Then I found the commit I was looking for very quickly. But maybe others don't use git history quite as extensively. I don't know. I'm only a documentation editor, but I've found it useful to browse history many times, often going back several years. When I first tangled with git I also found browsing history with gitk very elucidating - without an extensive example of a git repository in front of me it would have taken me even longer to master git. I would recommend using git clone with a shallow depth only by someone with a very poor internet connection or very limited disk space. Trevor ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Le vendredi 07 août 2009 à 21:00 -0700, Patrick McCarty a écrit : > Personally, I browse git history on the command line quite often when > working with LilyPond. So do I, or sometimes with gitk. There are already many Git features to master for daily use, I'm not keen on adding to the CG the shallow cloning option, which should be only used by people that want to look at the code or build snapshots of revisions between releases without contributing anything IMO ; every serious contributor may need to browse the entire history. Cheers, John signature.asc Description: Ceci est une partie de message numériquement signée ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
On Fri, Aug 07, 2009 at 08:39:16PM -0700, Mark Polesky wrote: > > Patrick McCarty wrote: > > > I've just tested git's shallow cloning feature. It's pretty neat. > > :-) > > > > From what I can see, shallow clones would be okay for *casual* > > contributors that are only sending patches based on the tip of master. > > > > However, since git history is limited to the depth of the clone, then > > shallow clones would not permit a developer to revert a commit from, > > say, three weeks ago. > > > > In other words, I think both the "git clone" and "git clone --depth" > > methods should be included in the CG. > > To me, it seems that a developer should be able to just stick to > shallow clones for everyday use. I assume that one could simply > increase the depth of the clone when needed. Is that true? I've never > needed to revert a three-week old commit, but if I needed to, I > figure I would just do another clone with a greater depth. > > Does it work that way? I suppose that sounds reasonable, but if you "reclone" a repository two or three times, each time with greater depth, then you'll probably be using more bandwidth than if you had just downloaded the entire repo initially (with "git clone"). Really, it all depends on how developers use git history. Personally, I browse git history on the command line quite often when working with LilyPond. A while back, I needed to find a change made to a certain file several years ago (I don't remember which file). To exacerbate the problem, the file had been renamed once or twice. But git makes this easy: $ git log -p --follow file.scm Then I found the commit I was looking for very quickly. But maybe others don't use git history quite as extensively. I don't know. Thanks, Patrick ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Patrick McCarty wrote: > I've just tested git's shallow cloning feature. It's pretty neat. > :-) > > From what I can see, shallow clones would be okay for *casual* > contributors that are only sending patches based on the tip of master. > > However, since git history is limited to the depth of the clone, then > shallow clones would not permit a developer to revert a commit from, > say, three weeks ago. > > In other words, I think both the "git clone" and "git clone --depth" > methods should be included in the CG. To me, it seems that a developer should be able to just stick to shallow clones for everyday use. I assume that one could simply increase the depth of the clone when needed. Is that true? I've never needed to revert a three-week old commit, but if I needed to, I figure I would just do another clone with a greater depth. Does it work that way? - Mark ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
On Mon, Aug 03, 2009 at 06:45:38PM -0700, Graham Percival wrote: > On Mon, Aug 03, 2009 at 06:18:20PM -0700, Mark Polesky wrote: > > > > Graham Percival wrote: > > > > > If you do restart, try the > > > git clone --depth 1 git://URL > > > method. (the CG will probably be updated to use this method in a > > > week or so) > > > > it seems that this method is not suitable for developers with > > push access. You can't push from it -- see below. > > I've done it. > > > --depth > > Create a shallow clone with a history truncated to the specified > > number of revisions. A shallow repository has a number of > > limitations (you cannot clone or fetch from it, nor push from nor > > into it), but is adequate if you are only interested in the recent > > history of a large project with a long history, and would want to > > send in fixes as patches. > > I read that before, but I tried it anyway. git push was just > fine. I was using git 1.5.6.5. I'll double-check in a few days, > before changing the CG. I've just tested git's shallow cloning feature. It's pretty neat. :-) >From what I can see, shallow clones would be okay for *casual* contributors that are only sending patches based on the tip of master. However, since git history is limited to the depth of the clone, then shallow clones would not permit a developer to revert a commit from, say, three weeks ago. In other words, I think both the "git clone" and "git clone --depth" methods should be included in the CG. Thanks, Patrick ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Trevor Daniels wrote: > Something must have happened between the last > time git worked and the first time it failed. > Anything you did during that time is a suspect. The only other things I can imagine: 1) my anti-virus software might have done something 2) my security might have been compromised by some outside attack No idea. - Mark ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Mark Polesky wrote Tuesday, August 04, 2009 9:23 AM Does anyone know -- is there any harm in just generating a new SSH key pair? No harm at all. I think my modem gives me a different IP address from time to time. It (well, your ISP) will - that's normal. I could be making that up, and I don't know if that would even affect the SSH stuff. It won't affect it at all. But I'm willing to try anything at this point. Would generating a new key pair cause problems? No, but I doubt it will fix anything either. Did rebuilding the repository have no effect? BTW your env variables seem normal, so no help from there, I'm afraid. Something must have happened between the last time git worked and the first time it failed. Anything you did during that time is a suspect. Trevor ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
On Tue, Aug 04, 2009 at 01:23:11AM -0700, Mark Polesky wrote: > Does anyone know -- is there any harm in just generating a new > SSH key pair? Absolutely no problems. Go ahead. > I think my modem gives me a different IP address > from time to time. I could be making that up, and I don't know > if that would even affect the SSH stuff. Most ISPs deliberately don't guarantee IP addresses, so it's quite reasonable that it would change. This won't effect SSH, though. But in the interest of problem elimination, definitely try generating a new key pair. Just remember that savannah needs an hour or so to update it. Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Does anyone know -- is there any harm in just generating a new SSH key pair? I think my modem gives me a different IP address from time to time. I could be making that up, and I don't know if that would even affect the SSH stuff. But I'm willing to try anything at this point. Would generating a new key pair cause problems? - Mark ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
On Mon, Aug 03, 2009 at 06:18:20PM -0700, Mark Polesky wrote: > > Graham Percival wrote: > > > If you do restart, try the > > git clone --depth 1 git://URL > > method. (the CG will probably be updated to use this method in a > > week or so) > > it seems that this method is not suitable for developers with > push access. You can't push from it -- see below. I've done it. > --depth > Create a shallow clone with a history truncated to the specified > number of revisions. A shallow repository has a number of > limitations (you cannot clone or fetch from it, nor push from nor > into it), but is adequate if you are only interested in the recent > history of a large project with a long history, and would want to > send in fixes as patches. I read that before, but I tried it anyway. git push was just fine. I was using git 1.5.6.5. I'll double-check in a few days, before changing the CG. In short: buggered if I can figure out how to use git. :/ Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Graham Percival wrote: > If you do restart, try the > git clone --depth 1 git://URL > method. (the CG will probably be updated to use this method in a > week or so) Graham, it seems that this method is not suitable for developers with push access. You can't push from it -- see below. - Mark --depth Create a shallow clone with a history truncated to the specified number of revisions. A shallow repository has a number of limitations (you cannot clone or fetch from it, nor push from nor into it), but is adequate if you are only interested in the recent history of a large project with a long history, and would want to send in fixes as patches. ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Mark Polesky wrote Monday, August 03, 2009 11:38 PM Trevor Daniels wrote: These commands will fail. The correct command is $ git reset --hard origin/master It gets up to 50%, then crashes: Checking out files: 50% (963/1926) (here the Windows error box pops up) Once I close the pop-up, the line changes to: fatal: Could not reset index file to revision 'origin/master'. When I try resetting master to an *earlier* commit in my own repo, it again crashes with the (git) error: child process exited abmormally I think deleting your repository and rebuilding it from Savannah is the next thing to try. It only takes 20 mins or so, and I don't know what else you can do. I don't see the connection of a faulty repo with some of your errors, but this problem with reset and your earlier one with validate seem to suggest a screwed repo. Trevor ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Trevor Daniels wrote: > These commands will fail. The correct command is > > $ git reset --hard origin/master It gets up to 50%, then crashes: Checking out files: 50% (963/1926) (here the Windows error box pops up) Once I close the pop-up, the line changes to: fatal: Could not reset index file to revision 'origin/master'. When I try resetting master to an *earlier* commit in my own repo, it again crashes with the (git) error: child process exited abmormally ? - Mark ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Mark Polesky wrote Monday, August 03, 2009 4:25 PM Graham Percival wrote: What does: git reset --hard origin or git reset --hard origin master do? I'd expect one of those to set you to a working state. (NB: by "I'd expect", I mean "as a user, I think the program should do this". Unfortunately, as somebody who's been fighting with git for years, I have no confidence that git /will/ behave in that manner) $ git reset --hard origin fatal: ambiguous argument 'origin': unknown revision or path not in the working tree. Use '--' to separate paths from revisions $ git reset --hard origin master fatal: ambiguous argument 'origin': unknown revision or path not in the working tree. Use '--' to separate paths from revisions These commands will fail. The correct command is $ git reset --hard origin/master Let me know if that works. Trevor ps I'm still unable to discern the cause of your problems; maybe the reset hard will help, although a few rogue commits are not likely to cause some of your other problems. ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Graham Percival wrote: > What does: > git reset --hard origin > or > git reset --hard origin master > do? I'd expect one of those to set you to a working state. (NB: > by "I'd expect", I mean "as a user, I think the program should do > this". Unfortunately, as somebody who's been fighting with git > for years, I have no confidence that git /will/ behave in that > manner) $ git reset --hard origin fatal: ambiguous argument 'origin': unknown revision or path not in the working tree. Use '--' to separate paths from revisions $ git reset --hard origin master fatal: ambiguous argument 'origin': unknown revision or path not in the working tree. Use '--' to separate paths from revisions It's comforting to know that I'm not the only one fighting git! - Mark ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
On Mon, Aug 03, 2009 at 08:54:58AM +0100, Trevor Daniels wrote: > > Mark Polesky wrote Monday, August 03, 2009 2:30 AM >> > 3. >> git gui/Repository/Verify Database >> With that, I get: >> Error: Command Failed >> git-fsck-objects.exe has encountered a problem and needs to close. >> We are sorry for the inconvenience. > > So there is no problem starting git tasks, > but all fail during processing. Errors 2 > and 3 don't need ssh, so that isn't the > problem. Agreed. What does: git reset --hard origin or git reset --hard origin master do? I'd expect one of those to set you to a working state. (NB: by "I'd expect", I mean "as a user, I think the program should do this". Unfortunately, as somebody who's been fighting with git for years, I have no confidence that git /will/ behave in that manner) >> I'm thinking of removing my entire lilypond repo and starting >> over from the beginning. If that doesn't work, I'll uninstall/ >> reinstall git. > > Well, that would eliminate one possibility, > and it's easy to do, but I'm not convinced > you have a problem with the git code. If you do restart, try the git clone --depth 1 git://URL method. (the CG will probably be updated to use this method in a week or so) Cheers, - Graham ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Mark Polesky wrote Monday, August 03, 2009 2:30 AM Mark Polesky wrote: > 199.232.41.69 ssh-rsa B3NzaC1yc2E ... > > It's the public key of the Savannah server. In > case you've lost it, I've attached one. My copy of the public key looks fine. Any other ideas? Not shots in the dark. It's difficult helping from a distance as there's all sorts of trivial things which would give more clues but I don't know what exactly you've done to investigate this, or what precisely your operating regimen is. Let's summarise: 1. $ git pull ssh://sv/srv/git/lilypond.git master fatal: The remote end hung up unexpectedly 2. when I try "update" through the gitk screen, I get a rather disturbing message... Error reading commits: child killed: unknown signal 3. git gui/Repository/Verify Database With that, I get: Error: Command Failed git-fsck-objects.exe has encountered a problem and needs to close. We are sorry for the inconvenience. 4. $ git fetch The authenticity of 'git.sv.gnu.org (199.232.41.69)' can't be established. So there is no problem starting git tasks, but all fail during processing. Errors 2 and 3 don't need ssh, so that isn't the problem. The git repo is a common factor, but I don't see how a problem there could generate error 4. The common problem seems to be in locating files under your account, ~/, from tasks running under the bash shell. What does the env command return? Is HOME still set correctly (should point to your ~/)? Perhaps you could mail me the output from env. Enter env > env.txt to get it into a file. I'm thinking of removing my entire lilypond repo and starting over from the beginning. If that doesn't work, I'll uninstall/ reinstall git. Well, that would eliminate one possibility, and it's easy to do, but I'm not convinced you have a problem with the git code. Trevor ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Mark Polesky wrote: > > 199.232.41.69 ssh-rsa B3NzaC1yc2E ... > > > > It's the public key of the Savannah server. In > > case you've lost it, I've attached one. > > My copy of the public key looks fine. Any other ideas? I'm thinking of removing my entire lilypond repo and starting over from the beginning. If that doesn't work, I'll uninstall/ reinstall git. Anyone I should e-mail before doing this? All the info I find online is either too hard for me to understand or just doesn't help. I feel like everyone else (on Linux) is working with a Swiss Army knife, and all I have (on Windows) is a rusty paper clip. - Mark ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Trevor Daniels wrote: > > $ git fetch > > The authenticity of 'git.sv.gnu.org (199.232.41.69)' can't be established. > > RSA key fingerprint is > > Are you sure you want to continue connecting (yes/no)? > > This means the known-hosts file has been deleted > or corrupted or can't be found. It should be in > your .ssh directory. It should contain one long > line beginning > > 199.232.41.69 ssh-rsa B3NzaC1yc2E ... > > It's the public key of the Savannah server. In > case you've lost it, I've attached one. My copy of the public key looks fine. - Mark ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Mark Polesky wrote Sunday, August 02, 2009 10:03 PM So I did git fetch and got this: $ git fetch The authenticity of 'git.sv.gnu.org (199.232.41.69)' can't be established. RSA key fingerprint is Are you sure you want to continue connecting (yes/no)? This means the known-hosts file has been deleted or corrupted or can't be found. It should be in your .ssh directory. It should contain one long line beginning 199.232.41.69 ssh-rsa B3NzaC1yc2E ... It's the public key of the Savannah server. In case you've lost it, I've attached one. and it doesn't matter how I respond. It always goes into an endless loop. I always have to force-quit. Do I need to take this to the git mailing list? I don't think you have a git problem. Several of the messages you've reported are from Windows. (But I may be wrong :) Trevor known_hosts Description: Binary data ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
So I did git fetch and got this: $ git fetch The authenticity of 'git.sv.gnu.org (199.232.41.69)' can't be established. RSA key fingerprint is Are you sure you want to continue connecting (yes/no)? and it doesn't matter how I respond. It always goes into an endless loop. I always have to force-quit. Do I need to take this to the git mailing list? - Mark ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Mark Polesky wrote Sunday, August 02, 2009 1:43 AM Trevor Daniels wrote: > Error reading commits: > child killed: unknown signal I've never seen this error message, but it suggests a problem with the git repo. Try verifying the database. git gui/Repository/Verify Database With that, I get: Error: Command Failed git-fsck-objects.exe has encountered a problem and needs to close. We are sorry for the inconvenience. This is an MS generic error message issued as a catch-all, so the underlying problem is not easy to find. Presumably you've tried the obvious re-boot? Also are you using the same account as before - maybe git fsck needs admin privileges. Have you installed any new software? Or you might get a clue from the 'Click here' link at the bottom of the message box. Look for the module name, then Google for that with something like 'error'. Trevor ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Trevor Daniels wrote: > > Error reading commits: > > child killed: unknown signal > > I've never seen this error message, but it > suggests a problem with the git repo. Try > verifying the database. git gui/Repository/Verify Database With that, I get: Error: Command Failed git-fsck-objects.exe has encountered a problem and needs to close. We are sorry for the inconvenience. > A shot in the dark - check your path hasn't > got screwed. Looks okay, but I don't really know what it should look like. Anyone else? I can't push or pull or nothing. Trying to stay calm... (: - Mark ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Mark Polesky wrote Saturday, August 01, 2009 9:09 PM Patrick McCarty wrote: > $ git pull ssh://sv/srv/git/lilypond.git master > fatal: The remote end hung up unexpectedly Well, that URI is kind of bogus. :-) It worked fine yesterday. It works fine here today too. That's not the problem. This should work: $ git pull git://git.sv.gnu.org/lilypond.git master No luck. Same fatal message. And when I try "update" through the gitk screen, I get a rather disturbing message... Error reading commits: child killed: unknown signal I've never seen this error message, but it suggests a problem with the git repo. Try verifying the database. git gui/Repository/Verify Database You'd think they could've come up with something a little less dark. Geez. :) When I tried this: $ git pull ssh://myn...@git.sv.gnu.org/srv/git/lilypond.git master I got something about unverified identity, then do I want to continue? Say no, and then the terminal just keeps printing the same message over and over until I force-quit. Any other ideas? A shot in the dark - check your path hasn't got screwed. Trevor ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
Patrick McCarty wrote: > > $ git pull ssh://sv/srv/git/lilypond.git master > > fatal: The remote end hung up unexpectedly > > Well, that URI is kind of bogus. :-) It worked fine yesterday. > This should work: > > $ git pull git://git.sv.gnu.org/lilypond.git master No luck. Same fatal message. And when I try "update" through the gitk screen, I get a rather disturbing message... Error reading commits: child killed: unknown signal You'd think they could've come up with something a little less dark. Geez. When I tried this: $ git pull ssh://myn...@git.sv.gnu.org/srv/git/lilypond.git master I got something about unverified identity, then do I want to continue? Say no, and then the terminal just keeps printing the same message over and over until I force-quit. Any other ideas? - Mark ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
Re: git hang-ups
On Sat, Aug 01, 2009 at 12:05:35PM -0700, Mark Polesky wrote: > > I'm getting this: > > $ git pull ssh://sv/srv/git/lilypond.git master > fatal: The remote end hung up unexpectedly Well, that URI is kind of bogus. :-) This should work: $ git pull git://git.sv.gnu.org/lilypond.git master ___ I can do $ git pull because I have my .git/config set up with this information: [remote "origin"] url = ssh://myn...@git.sv.gnu.org/srv/git/lilypond.git fetch = +refs/heads/*:refs/remotes/origin/* [branch "master"] remote = origin merge = refs/heads/master rebase = true HTH, Patrick ___ lilypond-devel mailing list lilypond-devel@gnu.org http://lists.gnu.org/mailman/listinfo/lilypond-devel
