2000-09-18 23:18:34 +04:00
|
|
|
I did not code a warnings system with all the warning pages (see
|
2000-11-28 03:24:42 +03:00
|
|
|
tinderbox1). I am not sure this belongs in Tinderbox proper. I think
|
|
|
|
that a separate webscript running a version of "lint" would be a
|
2001-03-15 20:57:20 +03:00
|
|
|
better choice. This web script would be a tinderbox "build". This
|
|
|
|
would totally decouple the warning system from the tinderbox error
|
|
|
|
parser and provide more robust warnings then a traditional
|
|
|
|
compiler. IF we need to put it in Tinderbox we should create a special
|
|
|
|
library which is called every time the error parser detects a warning.
|
2000-06-22 08:17:19 +04:00
|
|
|
|
2000-11-28 03:24:42 +03:00
|
|
|
I need a Javascript expert to look over the current versions of CVS
|
2000-09-18 23:18:34 +04:00
|
|
|
Blame and Tinderbox1 and ensure that I have the best features of these
|
|
|
|
javascript programs in tinderbox2/src/lib/HTMLPopUp/MozillaLayers.pm.
|
|
|
|
Please keep the code neat and nicely indented.
|
2000-06-22 08:17:19 +04:00
|
|
|
|
|
|
|
I need to allow the VC system to have a daemon gather the VC
|
|
|
|
information and pass it to tinderbox via the same update methods wich
|
|
|
|
Builds uses. Then both the tinder.cgi and this new cvs process could
|
|
|
|
be run out of cron periodically and the webpages would not be delayed
|
|
|
|
if CVS takes a long time to respond this might improve the response
|
|
|
|
time of the web updates. Also the OS process scheduler could do a
|
|
|
|
better job of scheduling if there were two different processes one
|
|
|
|
which was IO bound and one which was CPU bound.
|
|
|
|
|
2000-10-18 00:30:29 +04:00
|
|
|
It would be really nice if CVS could mail tinderbox the updates which
|
2000-06-22 08:17:19 +04:00
|
|
|
apply to the branches of interest, then tinderbox would not have to
|
|
|
|
pool and it would get the branch information which it needs.
|
2000-09-18 23:18:34 +04:00
|
|
|
|