> > I looked at this and it tells you how to get the sources via
> > CVSup or cvs, but not how to keep them in sync so to speek.
> > I would like the ability to make my own branches and so forth
> > to play with certain aspects of the database. However, I would
> > like the ability to fold in patches in a big group without having
> > to apply each patch by hand.
> > Any other ideas?
The way I work is to use CVSup to check out versions locally on my
machine. Then, when I change a file I save a version as filename.orig
and these changed files tend to accumulate. When I'm ready to submit a
patch, I run a little utility I wrote which traverses the development
tree and generates a patch file for each file/file.orig pair, putting
all of the files in a separate patch directory. I've got another utility
which can apply this set of patch files to the cvs source tree at
postgresql.org. Afterwards, I'll clean up by removing the .orig files
and doing a CVSup refresh then a cvs update on my tree.
If I'm doing a longer-term project, I'll freeze my tree and check out a
new one when I need to test with a current development tree.
Bruce has a utility he included in the distribution which generates a
single patch file for all file/file.orig pairs (afaik). Look in the
src/tools directory. I don't use that because I quite often have more
than one set of patches going at a time so will delete the patch files
which aren't relevant before submitting or committing them.
> And John Polstra tends to be very responsive to questions about his
> software :)
Very much so. There is a new rpm distribution of Modula-3 and John has
generated a new version of CVSup which is tailored to Linux (just
slightly different from his main FreeBSD version). I've been sitting on
the new code, intending to generate new static binaries for glibc2 and
for libc5. Will post on postgresql.org when they are ready.
- Tom
