drinkypoo's Journal: How do you resume a CVS checkout?

I'm using a modem and trying to check out some CVS projects. When svn fails I can just cd into the directory and type "svn up" and it picks right up where it left off. When I do this with cvs, it only updates successfully-downloaded files. Trying to issue the same checkout command from the same working directory as the first time results in EXTREME SLOWNESS as cvs tells me to move aside existing files. How do I resume a cvs checkout in a reasonable fashion? Do I really have to start all over? Because some projects are literally impossible for me to download in a single pass.

Re: How do you resume a CVS checkout?

[Em Adespoton]

I thought this was one of the reasons svn was created in the first place... cvs just doesn't support as much.

The only suggestion I can make is to check out the project in multiple passes, not all at once. This should work OK for you, even if it's more inconvenient.

Re:

[drinkypoo]

Oucheroo. There MUST be a script out there someplace to do it, comparing a ls to a remote list. Right? I mean, there has to be?

For a moment there....

[unitron]

...I thought you were talking about standing in line at the drug store cash register. :-)

(I liked it better when their name was Revco)

Re:

[drinkypoo]

I wasn't, but I do get a huge kick out of seeing "CVS/pharmacy" on the side of a building. If they remodeled they could temporarily hang a sign that said "pharmacy/.cvs". (or perhaps autotools/bootstrap.sh)

Log file quickly

[tqft]

Don't try anything yet.

I regularly broke my firefox cvs tree by having updates broken when on dial up. Delete the directory where it stopped and do cvs update - the log file should tell you where it failed.

Other than that it is redo the whole tree (deleting the directory where it stopped)

for posterity:

[drinkypoo]

wow, git sucks too. Anyway, cvsnt seems to have a slightly more rational client than old-schooly cvs. For what it's worth. It's still a bear, though. CVS at freedesktop.org is so bad that I would often have failures checking out a single source file (.c, .h, whatever) by name. Now they're going to git which forces the initial clone of a remote repo as a single action (it's better about subsequent failures, though.) All hail SVN, which sucks least!