[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Convenient way to commit?

From: Hans Schwaebli
Subject: Re: Convenient way to commit?
Date: Wed, 21 Mar 2007 03:23:02 -0700 (PDT)

so there is no out of the box script or command I just can use when I want do do this. Now I know. As I said, Eclipse IDE does allow me to do this *high level* commit (including recursive cvs add).
You ask why I want it for daily builds. They are run unattended. Sometimes at night. Then I am not there to tell which file to add to cvs and which not... see? And I don't know exactly which files are created during the build process. Maybe someone adds a new application component, that would be an additional jar file generated.
So I have to zip all this stuff and to commit this one large file. But by this the revision history for the artifact elements cannot be tracked anymore because they are all in that zip file. So checking in everything which is in a certain folder would be better for this from my understanding. The only thing which would be needed to do this in a productive way is to provide a recursive add command.
This file grows very fast to 2 GB in size. Linux cannot handle this large size and the cvs commit command fails unexpectedly. We then delete this file on the CVS file system from time to time. Not a nice thing. Besides that, CVS needs a lot of time to handle such large files when synchronizing, tagging and checking out.
Spiro Trikaliotis <address@hidden> wrote:

* On Wed, Mar 21, 2007 at 02:00:03AM -0700 Hans Schwaebli wrote:

> I want to commit all files folders and subfolders of a certain folder,
> may or may they not already have been added to CVS.

This might be a bad idea in many cases. Compare your question about
ignoring files which once were in CVS.

I think it is much better to decide yourself, by hand, what you want to
add/delete, and what not.

> I want to automate this. So I need a foolproof way. But there are no
> convenience commands of CVS which allow me this. Or is there a way?

You can use:

1. Use the output of "cvs -nq up" to find the directories which have been
You can do this via a shell skript.

2. Now, add all directories which have been added with "cvs add". xargs
might help you here.

3. Use the output of "cvs -nq up" to find the files which have been
added, and find the files which have been removed.
You can do this via a shell skript.

4. Now, add all files which have been added with "cvs add", and remove
all files which have been removed with "cvs remove". Again, xargs
will help you.

5. Check if everything is right "cvs -nq up", and commit it with "cvs

I once have done it myself, and thus, I know it works this way.

> Why not just writing cvs -f commit foldername? If a resource is not
> added, the command would add it automatically and if a resource is
> deleted by the commit, it is deleted automatically.

Because you might end up with files in CVS you do not want to have
there. Compare your other post, again.

> Such things are badly needed for stuff like Continuous Builds and
> Nightly Builds.

Could you please tell me why you would need this for these? I do not see
any reason.

Best regards,

Spiro R. Trikaliotis

info-cvs mailing list

TV dinner still cooling?
Check out "Tonight's Picks" on Yahoo! TV.

No need to miss a message. Get email on-the-go
with Yahoo! Mail for Mobile. Get started.
reply via email to

[Prev in Thread] Current Thread [Next in Thread]