Re: [jcvs] problem with Entries file

[Matthias Tichy <mtt@uni-paderborn.de>]

hi,

i could reproduce the problem. If a merge on an update operation 
succeded (normal text file merge, or binary file conflict) the 
CVSProject sets the timestamp on the entry to "Result of merge".
But the setTimestamp()-Method in CVSEntry does overwrite this "Result
of merge" string (which is correctly generated by the normal cvsclient 
by the way) with the empty string "".

CVSEntry:
     else if ( tstamp.equals( "Result of merge" ) )
          {
               // This file was "merged" timestamp must show modified
               this.timestamp = "";
               this.tsCache = null;
          }

As an result, the line in the CVS/Entries files has an empty 
conflict/timestamp field.

Is the above assignment correct? Or is it a bug?

I changed the code in a way, that "Result of merge" is written to the 
Entries-file. At the moment it seems to cause no problems, but i would 
be nice to hear a statement from a person, who knows jcvs better than i.

Matthias


> Hi Matthias,
> 
> Any idea how that entry was created in the first place? Did the command
> line CVS program write those entries, or did jCVS?
> 
> tim.
> 
> 
>>hi,
>>
>>i experienced the following problem that the Entries file contained 
>>entries without a timestamp. For example
>>
>>/test.java/1.1///
>>
>>These entries cause the problem, that the files are considered changed 
>>and the pserver tries to merge them indefenitly.
>>
>>Has anybody experienced the same problem or can enyone give me a hint 
>>how such a wrong Entries file can be created.
>>
>>Matthias
>>
>>-- 
>>            mbm
>>UP THE IRONS!
>>
>>--------------------------------------------------------------------
>>View List Archive:   <http://www.gjt.org/mlist/jcvs/threads.html>
>>Search List Archive: <http://csdl.ics.hawaii.edu:8100/>
>>To unsubscribe from this list:
>>   <http://www.gjt.org/servlets/MailingLists/UnsubForm.html/jcvs>
>>
>