I know nothing of cvs administration, but would it be possible (without too much effort) to add annotation to the web cvs interface? I find that knowing the last time a particular piece of code changed can make regression testing and assorted other tasks happen a lot faster.
Compare: annotated file:
http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/tiny-cobol/development/Makefi... raw file:
http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/tiny-cobol/development/Makefi...
-- Jeff S
_________________________________________________________________ The new MSN 8: smart spam protection and 2 months FREE* http://join.msn.com/?page=features/junkmail
Jeff Smith wrote:
I know nothing of cvs administration, but would it be possible (without too much effort) to add annotation to the web cvs interface? I find that knowing the last time a particular piece of code changed can make regression testing and assorted other tasks happen a lot faster.
from the docs, it seems we need to have the user running the httpd server have rw to the CVSROOT/history directory... that shouldn't be too hard to do...
just a matter of time... I've added a bug request for this (#1214)
A+
I'm not sure that's right, and I am sure that this is very unwanted.
The reason I don't think this is right is that the rw access is required in order to create a lock file. Allowing it will have several very bad side effects.
* Anyone who breaks into the web site will have instant write access to the CVS - VERY BAD!!! * The chances that, when Alexander runs "cvs ci", he will have to wait for CVS locks to be cleared.
Instead, running cvs annotate with the -n switch (i.e. - "cvs -n annotate filename") doesn't create a lock file, and will probably not require rw access.
Shachar
Eric Pouech wrote:
Jeff Smith wrote:
I know nothing of cvs administration, but would it be possible (without too much effort) to add annotation to the web cvs interface? I find that knowing the last time a particular piece of code changed can make regression testing and assorted other tasks happen a lot faster.
from the docs, it seems we need to have the user running the httpd server have rw to the CVSROOT/history directory... that shouldn't be too hard to do...
just a matter of time... I've added a bug request for this (#1214)
A+
Shachar Shemesh wrote:
I'm not sure that's right, and I am sure that this is very unwanted.
The reason I don't think this is right is that the rw access is required in order to create a lock file. Allowing it will have several very bad side effects.
- Anyone who breaks into the web site will have instant write access to the CVS - VERY BAD!!!
- The chances that, when Alexander runs "cvs ci", he will have to wait for CVS locks to be cleared.
Instead, running cvs annotate with the -n switch (i.e. - "cvs -n annotate filename") doesn't create a lock file, and will probably not require rw access.
sure, the main issue we have is that winehq runs cvsweb as the web cvs frontend, and cvsweb only supports annotation under the condition I posted (IOW, it doesn't implement the annotate feature - on WWW side - with cvs annotate). But you're right on the constraints we get.
so either check if a newer cvsweb implements annotate with cvs annotate, or we hack cvsweb to get what we want/need...
A+
* Eric Pouech (eric.pouech@wanadoo.fr) wrote:
sure, the main issue we have is that winehq runs cvsweb as the web cvs frontend, and cvsweb only supports annotation under the condition I posted (IOW, it doesn't implement the annotate feature - on WWW side - with cvs annotate). But you're right on the constraints we get.
so either check if a newer cvsweb implements annotate with cvs annotate, or we hack cvsweb to get what we want/need...
There's all sorts of versions of cvsweb floating around, no doubt one of them contains what you want.
BTW: w.r.t. rw permissions on CVS for annotation functionality. Do you run cvsweb (or whatever) on a slave *copy* of the CVS repository rather than leaving any possible window of opportunity for the master repository to get hit?
Cheers, Geoff
Geoff Thorpe wrote:
- Eric Pouech (eric.pouech@wanadoo.fr) wrote:
sure, the main issue we have is that winehq runs cvsweb as the web cvs frontend, and cvsweb only supports annotation under the condition I posted (IOW, it doesn't implement the annotate feature - on WWW side - with cvs annotate). But you're right on the constraints we get.
so either check if a newer cvsweb implements annotate with cvs annotate, or we hack cvsweb to get what we want/need...
There's all sorts of versions of cvsweb floating around, no doubt one of them contains what you want.
BTW: w.r.t. rw permissions on CVS for annotation functionality. Do you run cvsweb (or whatever) on a slave *copy* of the CVS repository rather than leaving any possible window of opportunity for the master repository to get hit?
I dunno. I lost track of it when the server got moved to CW. Jeremy Newman does know. A+
-
Eric Pouech -
Geoff Thorpe -
Jeff Smith -
Shachar Shemesh