Re: Trouble with cvs -d :pserver:[email protected]:/cvs -f -z3 update -PdA texi2html

Dumas Patrice wrote:

>>Oops.  Yep.  That's probably it.  I should have spotted it.  I'll get in 
>>and rename it on the server.
>>
>>Pat or anyone else who has the directory checked out may see errors on 
>>their next update.  You can either grab a fresh checkout or just rm -rf 
>>your Tests/cvs directory and rename the entry in Tests/CVS/Entries from 
>>cvs to ccvs, or delete the entry entirely from the Entries file, and 
>>then update.
>>    
>>
>
>I tried that removing the entry in Tests/CVS/Entries, and do a cvs up
>in Tests. But the ccvs directory didn't appear.
>

You need to specify `-d' to `cvs update'.  I usually have it in my 
~/.cvsrc, so I forgot to mention it.  Sorry about that.  I usually also 
put `-P' in there (as an argument for update), but your ccvs_res 
directory won't come out if you do that.  `-P' instructs CVS to prune 
empty directories, useful for ignoring directories all of the files have 
been removed from, but trouble sometimes when you are just adding a new 
empoty directory.  My usual work around is to leave `-P' in my ~/.cvsrc 
and immediately add some file (for instance, `.cvsignore' or 
`Makefile.am') to the directory after creation.

> And then I made
>$ cvs co .
>But it was a very bad idea, now I have a lot of projects checked out
>in the Tests directory (activitymail, *all public projects in domain,
>autobuilder, cms2cvs, cmtools, cvsfororacleforms and a lot more)
>:-(
>
>I have checked out a fresh copy too. I think I will just release the old 
>one (after commiting things that have to) if you think it is the right
>thing to do.
>

Sorry about that.  You shouldn't have any problem committing as long as 
the change isn't in the old `cvs' directory.

A fresh checkout does sound easiest.

>>Someone will probably have to edit the Makefile.am with the new path 
>>too.  I'll do it later if noone else has time.
>>    
>>
>
>I did it. One more thing remains to be done: change cvs_res directory
>into ccvs_res. Derek, could you do that, please ?
>

Done.  Thanks for the Makefile.am change.

>$ cvs -d :pserver:[email protected]:/cvs co texi2html
>
>and things didn't went through the tunnel. Then I did (I believe this is right)
>
>$ cvs -d :pserver:pertusus@localhost:/cvs login
>$ cvs -d :pserver:pertusus@localhost:/cvs checkout texi2html
>
>and things went through the tunnel. It is possible that I missed this 
>information on the site.
>

You seem to be correct.  The tunnel information neglects to explain why, 
exactly, the user set up the tunnel and how to use it.  CollabNet is 
supposed to upgrade the SourceCast version on cvshome.org within the 
next few months, so I might wait until then before I try and rewrite 
their documentation.  We're adopting new developers at a slow enough 
rate that I can probably remember to tell them myself for now.

>Maybe it should also be mentionned that the tunnel won't work if the
>client is allready a cvs server (that's a guess, I haven't tried, though).
>

That's true.  I don't think it is mentioned on the site, but if you are 
using a version 1.11.1p1 or later of CVS, you can tunnel some other port 
(I use 21010) on the local side using ssh as long as the destination 
port on cvshome.org is 2401.  Then the CVSROOT spec you'd use would be 
`:pserver:pertusus@localhost:<port>/cvs', where <port> is the port you 
tunnelled from localhost.  Again, for my example, that's 
`:pserver:dprice@localhost:21010/cvs'.

>And last the server could refuse some cvs commands if they are not issued
>from localhost (commit, login...), and it would even be better if the server
>refuse the connection before the password is send.
>  
>

That might be nice.  The version of CVS running on cvshome.org would 
need to be hacked to support it, though.  I'll think about talking to 
Collab after they upgrade SourceCast.

Derek

-- 
                *8^)

Email: [email protected]

Get CVS support at <http://ximbiot.com>!
-- 
I've never made a mistake in my life.  I thought I had once, but it turned
out that I hadn't.