CVS upgrade

2003-02-10 Thread Amit Sharma
Hi All,
We are using CVS 1.11 Server on Suse Linux Enterprise Edition 7.0 &
want to upgrade it to CVS 1.11.3 Please let us know if anybody using the
same configuration have found any stability issue with this version.

Regards,
Amit.



___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs



How to add files to a Branch

2003-02-10 Thread Fabio Fracassi
Hi, 

I have some problems in updateing my current branch.
The repository has two branches, the HEAD one, and a branch called MERGE, 
where i did a merge. There where a couple of tags on the HEAD (like VER1, 
VER2, ...), and any time I completed and successfully tested a merge on the 
branch, I tagged the branch something like VER1_MERGED.

Now sadly the HEAD branch was incomplete, and I had to do the last merge step 
manually, which lead to several files which are not in the repository, and 
the cvs-information in the workdir might be inconsistent. 

My question is how can I update the MERGE branch with the latest version I 
have?

if I do a cvs add it says:
cvs add: cannot add file on non-branch tag VER2_MERGED

TIA

Fabio





___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs



User variable (Expert question)

2003-02-10 Thread Sylvain Lemieux
Hi all,

I need the help of a CVS expert that could help solving the following
problem:

I have built a program to check commit and have people provide an
acceptance code when they commit there modification.
 
To do so, I am using a user variable in the commitinfo file to pass the
extra information. Here is the line I have in my commitinfo.
 
DEFAULT /serv/CVSROOT/commitcheck ${=COMMITCODE}

If I use the following command line:

cvs -s COMMITCODE=12345 commit

Then ${=COMMITCODE} will be replaced by 12345 and passed as an extra
argument.

However, if people do cvs commit, the variable is left undefined and my
commitcheck program isn't called, which create a flaw in my attempt to
restrict commit. This is real problem for me as people are exploiting
that flaw to bypass the commit security.
 
Is there a way to ensure my program is call even is the variable is
undefined?
 
Any help will be appreciated!

--
Sylvain Lemieux, ing.
PCI Geomatics Inc. (The french connection)
On ne peut tomber que si l'on essaie de marcher
Every rain cloud has a silver lining



___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs



Re: User variable problem

2003-02-10 Thread CHARLES HART, BLOOMBERG/ 499 PARK
There's a shell setting that makes it complain when you use an undefined value.
(But I don't recall what it is).  If you are using ksh, you can use
${parameter:-word} to substitute "word" in place of "parameter" if parameter
isn't defined.

DEFAULT /serv/CVSROOT/commitcheck ${COMMITCODE:-JUNK}

Then, in your script, if $1 == Junk, then you know the user left it off.

-CTH





___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs



checkout problem: UNRESOLVED

2003-02-10 Thread pgoverno
Hi 

I'am starting to use CVS 1.11.4 in a Windows 2000 machine, and it works fine, but i 
can't checkout a module. 
Example: 
Supose i have a subdir of CVSROOT like /XXX/YYY, if i make "CVS checkout /XXX/YYY" 
it works. 

But, if i have a line like "test/XXX/YYY" in my modules file and i make "CVS 
checkout test" i receive a message in terms of 
"cvs checkout: cannot find module 'test' - ignored ". 

Question: where is the mistake ? 

Thanks on advance 

PG 


___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs



Re: checkout problem: UNRESOLVED

2003-02-10 Thread Riechers, Matthew W
[EMAIL PROTECTED] wrote:
> 
> But, if i have a line like "test/XXX/YYY" in my modules file
> and i make "CVS checkout test" i receive a message in terms of
> "cvs checkout: cannot find module 'test' - ignored ".
> 
> Question: where is the mistake ?

Probably in the CVSROOT/modules file. First, make sure you commited the
change to the file. Does "test" show up when you tell CVS to dump the
module database 'cvs co -c'?

-Matt


___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs



Re: User variable (Expert question)

2003-02-10 Thread Mark D. Baushke
Sylvain Lemieux <[EMAIL PROTECTED]> writes:

> I need the help of a CVS expert that could help solving the following
> problem:
> 
> I have built a program to check commit and have people provide an
> acceptance code when they commit there modification.
>  
> To do so, I am using a user variable in the commitinfo file to pass the
> extra information. Here is the line I have in my commitinfo.
>  
> DEFAULT /serv/CVSROOT/commitcheck ${=COMMITCODE}
> 
> If I use the following command line:
> 
> cvs -s COMMITCODE=12345 commit
> 
> Then ${=COMMITCODE} will be replaced by 12345 and passed as an extra
> argument.
> 
> However, if people do cvs commit, the variable is left undefined and my
> commitcheck program isn't called, which create a flaw in my attempt to
> restrict commit. This is real problem for me as people are exploiting
> that flaw to bypass the commit security.
>  
> Is there a way to ensure my program is call even is the variable is
> undefined?
>  
> Any help will be appreciated!

First, a cvs import will not trigger the commitinfo checks, so there is
still a potential hole in your commit process.

Second, using cvs 1.11.5 and the ccvs version I see the following:

  % cvs ci -m test test.c
  cvs server: commitinfo:16: no such user variable ${=COMMITCODE}
  cvs server: Pre-commit check failed
  cvs [server aborted]: correct above errors first!
  % cvs -s COMMITCODE=12345 ci -m test test.c
  RCS file: /tmp/cvs-sanity/cvsroot/CVSROOT/test.c,v
  done
  Checking in test.c;
  /tmp/cvs-sanity/cvsroot/CVSROOT/test.c,v  <--  test.c
  initial revision: 1.1
  done
  cvs server: Rebuilding administrative file database
  % cat -n commitinfo
   1  # The "commitinfo" file is used to control pre-commit checks.
   2  # The filter on the right is invoked with the repository and a list 
   3  # of files to check.  A non-zero exit of the filter program will 
   4  # cause the commit to be aborted.
   5  #
   6  # The first entry on a line is a regular expression which is tested
   7  # against the directory that the change is being committed to, relative
   8  # to the $CVSROOT.  For the first match that is found, then the remainder
   9  # of the line is the name of the filter to run.
  10  #
  11  # If the repository name does not match any of the regular expressions in 
this
  12  # file, the "DEFAULT" line is used, if it is specified.
  13  #
  14  # If the name "ALL" appears as a regular expression it is always used
  15  # in addition to the first matching regex or "DEFAULT".
  16  DEFAULT $CVSROOT/CVSROOT/commitcheck foo=${=COMMITCODE}
  % cat /tmp/cvs-santiy/cvsroot/CVSROOT/commitcheck
  #!/bin/sh
  
  case $1 in
foo=12345) exit 0 ;;
*) exit 1 ;;
  esac
  %

So, I am unable to reproduce your problem. Perhaps you could try with a
newer version of cvs and see if you still have the problem?

Enjoy!
-- Mark


___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs



RE: User variable (Expert question)

2003-02-10 Thread Sylvain Lemieux
Mark,

I have made few more tests in an attempt to figure out the differences
between my setup and yours. The first one that poped to me is the fact
that my server sits 500Km from my office and runs on a Linux machine. My
first idea was to say that it shouldn't be matter, but it seems that it
does matter. I setup a local server on my machine and got the same error
you shown me in this email using both 1.10 and 1.11.5. (Pre-commit check
failed) However, when I access my real server, the one that is remote, I
get the following error:

cvs server: commitinfo:16: no such user variable ${=COMMITCODE}
Checking in test.txt;
/serv/cvsroot/v700/src/cv/cvbase/test.txt,v  <--  test.txt
new revision: 1.4; previous revision: 1.3
done

Please note that my server still uses cvs 1.10.6 which may be part of
the problem. I will get on it right now and get 1.11.5 running on it to
make further test.

Meanwhile, Can you see any reason why the server would react differently
as opposed to local configuration in this case?

Thanks for spending time on my problem.

--
Sylvain Lemieux, ing.
PCI Geomatics Inc. (The french connection)
On ne peut tomber que si l'on essaie de marcher
Every rain cloud has a silver lining


-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]] On Behalf Of
Mark D. Baushke
Sent: Monday, February 10, 2003 1:01 PM
To: Sylvain Lemieux
Cc: [EMAIL PROTECTED]
Subject: Re: User variable (Expert question) 

Sylvain Lemieux <[EMAIL PROTECTED]> writes:

> I need the help of a CVS expert that could help solving the following
> problem:
> 
> I have built a program to check commit and have people provide an
> acceptance code when they commit there modification.
>  
> To do so, I am using a user variable in the commitinfo file to pass
the
> extra information. Here is the line I have in my commitinfo.
>  
> DEFAULT /serv/CVSROOT/commitcheck ${=COMMITCODE}
> 
> If I use the following command line:
> 
> cvs -s COMMITCODE=12345 commit
> 
> Then ${=COMMITCODE} will be replaced by 12345 and passed as an extra
> argument.
> 
> However, if people do cvs commit, the variable is left undefined and
my
> commitcheck program isn't called, which create a flaw in my attempt to
> restrict commit. This is real problem for me as people are exploiting
> that flaw to bypass the commit security.
>  
> Is there a way to ensure my program is call even is the variable is
> undefined?
>  
> Any help will be appreciated!

First, a cvs import will not trigger the commitinfo checks, so there is
still a potential hole in your commit process.

Second, using cvs 1.11.5 and the ccvs version I see the following:

  % cvs ci -m test test.c
  cvs server: commitinfo:16: no such user variable ${=COMMITCODE}
  cvs server: Pre-commit check failed
  cvs [server aborted]: correct above errors first!
  % cvs -s COMMITCODE=12345 ci -m test test.c
  RCS file: /tmp/cvs-sanity/cvsroot/CVSROOT/test.c,v
  done
  Checking in test.c;
  /tmp/cvs-sanity/cvsroot/CVSROOT/test.c,v  <--  test.c
  initial revision: 1.1
  done
  cvs server: Rebuilding administrative file database
  % cat -n commitinfo
   1  # The "commitinfo" file is used to control pre-commit checks.
   2  # The filter on the right is invoked with the repository and a
list 
   3  # of files to check.  A non-zero exit of the filter program
will 
   4  # cause the commit to be aborted.
   5  #
   6  # The first entry on a line is a regular expression which is
tested
   7  # against the directory that the change is being committed to,
relative
   8  # to the $CVSROOT.  For the first match that is found, then
the remainder
   9  # of the line is the name of the filter to run.
  10  #
  11  # If the repository name does not match any of the regular
expressions in this
  12  # file, the "DEFAULT" line is used, if it is specified.
  13  #
  14  # If the name "ALL" appears as a regular expression it is
always used
  15  # in addition to the first matching regex or "DEFAULT".
  16  DEFAULT $CVSROOT/CVSROOT/commitcheck foo=${=COMMITCODE}
  % cat /tmp/cvs-santiy/cvsroot/CVSROOT/commitcheck
  #!/bin/sh
  
  case $1 in
foo=12345) exit 0 ;;
*) exit 1 ;;
  esac
  %

So, I am unable to reproduce your problem. Perhaps you could try with a
newer version of cvs and see if you still have the problem?

Enjoy!
-- Mark


___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs



___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs



checkout previous version with only version info

2003-02-10 Thread Mona Wong-Barnum
Hi:

I would like to know if there is a way to checkout a previous version of 
a file specifying *only* the version information.

I know I can checkout previous tagged version, but my previous versions 
are not tagged.  I know I can checkout previous version based on date but I have 
many on the same date (it's a long story).

I'm using cvs 1.11.5.  I've checked the info-cvs archive and didn't see 
the answer to the question.

All help is appreciated.

Sincerely,
Mona

==
Mona Wong-Barnum
National Center for Microscopy and Imaging Research
University of California, San Diego
http://ncmir.ucsd.edu/

"While you are busy dancing for others, just remember that
others are also busy dancing for you"
(A Landmark Instructor)
==




___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs



How to admin deep hierachy directories

2003-02-10 Thread Charles Sun
I have a tool that generates many new files and folders in my working area.
These files shall be CVS controlled.

For example, my working area looks like:

Project_folder --- folderA --- subfolderA1 --- fileA1.1
-- folderB --- subfolderB1 --- fileB1.1

After running my tool, it becomes:

Project_folder --- folderA --- subfolderA1 --- fileA1.1(updated)
-- subfolderA2 --- fileA2.1
-- folderB --- subfolderB1 --- fileB1.1
-- fileB1.2
-- folderC --- fileC.1

I don't want to find and add each new files and folders one by one. What is
the easiest way to maintain my working area?


I want to add a symbolic link in my working area.  Can CVS keep this and
distribute it to other co-developers?


Charles



___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs



RE: unknown compression

2003-02-10 Thread Amit Sharma
Try upgrading to latest CVS Versions, It has latter version of zlib
compression utility and hope it solves the problem

regards,
Amit


-Original Message-
From: Fabian Cenedese [mailto:[EMAIL PROTECTED]]
Sent: Thursday, January 23, 2003 1:49 PM
To: '[EMAIL PROTECTED]'
Subject: Re: unknown compression



> I am getting this error while checking out data from Linux CVS
>Server thru my Windows CVSClient. Can anybody help me to understand the
>cause for this problem.
>COMMAND supplied:
>cvs -z9 -q update test.h (in directory C:\work)
>ERROR:
>cvs update: inflate: unknown compression method
>cvs [update aborted]: reading from server: Input/output error
>Waiting for ur contribution.

For some reason it doesn't know compression. Try if it works without
the -z9 option. What cvs program and version are you using?

bye  Fabi





___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs


___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs



Re: checkout previous version with only version info

2003-02-10 Thread Mark D. Baushke
Mona Wong-Barnum <[EMAIL PROTECTED]> writes:

>   I would like to know if there is a way to checkout a previous
> version of a file specifying *only* the version information.

Yes. Just use the version number in place of the tag. For example:

  % cvs -d :pserver:[EMAIL PROTECTED]:/cvs login
  (Logging in to [EMAIL PROTECTED])
  CVS password: 
  % cvs -d :pserver:[EMAIL PROTECTED]:/cvs checkout -r1.129 \
ccvs/doc/cvsclient.texi
  U ccvs/doc/cvsclient.texi
  % cd ccvs/doc
  % cvs status cvsclient.texi 
  ===
  File: cvsclient.texi  Status: Up-to-date
  
 Working revision:  1.129
 Repository revision:   1.129   /cvs/ccvs/doc/cvsclient.texi,v
 Sticky Tag:1.129
 Sticky Date:   (none)
 Sticky Options:(none)

  %

Then, to get rid of the sticky tag, you would use the -A switch like:

  % cvs up -A cvsclient.texi 
  P cvsclient.texi
  % cvs status cvsclient.texi 
  ===
  File: cvsclient.texi  Status: Up-to-date
  
 Working revision:  1.130
 Repository revision:   1.130   /cvs/ccvs/doc/cvsclient.texi,v
 Sticky Tag:(none)
 Sticky Date:   (none)
 Sticky Options:(none)

  %

Note that you may also just "print" the version to a file with

  % cvs up -p -r1.129 cvsclient.texi > cvsclient.texi.1.129

if you found it useful to have an older version around for some reason.

Enjoy!
-- Mark


___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs



Re: checkout previous version with only version info

2003-02-10 Thread Fabian Cenedese

>I would like to know if there is a way to checkout a previous version of 
>a file specifying *only* the version information.
>
>I know I can checkout previous tagged version, but my previous versions 
>are not tagged.  I know I can checkout previous version based on date but I 

You can use exactly the same command, just use the version number instead
of the tag, so 

cvs co -r 1.5 file

bye  Fabi




___
Info-cvs mailing list
[EMAIL PROTECTED]
http://mail.gnu.org/mailman/listinfo/info-cvs