[cvsnt] Re: RFE:Parameter similar to "-f" on update command
lists at connectionbrazil.com
Tue Aug 23 11:14:12 BST 2005
Henrik Wahlberg wrote:
>> I can see what you're doing here, but I probably would use a tag IFS_711_A
>> for the /complete/ set of files of release IFS_711 for customer A. I'm not
>> a fan of partial tagging...
> We have been discussing that, and that would have to negative sideeffects:
> 1) it would be impossible to review/list those files specially
> modified for costumer 'A'
> 2) When IFS_712 is released, I'd have some tagging problems,
> detecting where to place the IFS_712_A tag. I cant use IFS_711A, as it's
> definately wrong on base files (those not modified for that costumer).
> But I could update to stickytag IFS_712 and then tag all files as
> IFS_712_A, then stickytag to Costumer_A (removing all other files on the
> sandbox) and move IFS_712_A on these files. then I should have a
> useable combination of IFS_712 and Costumer_A so we we can start our
> automated build.
Hm... wouldn't this be a case for a Customer_A /branch/ rather than tag? It
seems you have a large set of base files with a few configuration-type
files for specific customers. You'd just merge in the IFS_712 main line
code into the Customer_A branch, and given that you don't touch the large
majority of files on the branch, you get an easy merge.
You'd still have a difficulty seeing which files are modified on the
branch. There was a discussion some time ago when someone wanted to move
the branching point; this would be another case where this feature could
> This is a bit akward though, as we are talking 10 modules containing 25
> to 890 files each, in all 1825 files on a rough count.
> That would indeed massage the CVSNT server on each automated build. :-)
Don't forget the power of shell scripts/batch files running at night. There
are many things that seem to be missing in cvsnt that can be done
relatively easily with a few lines of script. Like your scenario.
More information about the cvsnt