Greg Stein wrote:
> Why do some names use text_base and others use pristine? Aren't those
> the same thing?
The ones using 'pristine' are the ones added by Hyrum's SVN_EXPERIMENTAL
code, which, for the moment, run in parallel with the old ones, and then
later will replace the old ones. I haven't settled on which name is
more suitable for the final version, but it should be just one not a
mixture of both. At the moment, the names occur in pairs like these:
{ text_base_abspath, pristine_abspath } are the paths to two separate
copies of the same content;
{ text_base_checksum_md5, pristine_checksum_sha1 } are two different
kinds of checksum for the same content; we will probably need to keep
both of them.
So you're right, they mean the same thing logically, and only have this
temporary distinction of WC-1 vs. WC-NG. In the interest of naming
clarity, I'll think of a better way. Maybe:
{ text_base_abspath, pristine_abspath } - these two will be resolved
to just one very soon so I'll leave this dual naming for now;
{ text_base_checksum_md5, pristine_checksum_sha1 } - I'll modify the
patch before committing it, to make both of these say 'text_base' (or
perhaps both say 'pristine').
Good?
- Julian
> On Wed, Apr 7, 2010 at 13:05, Julian Foad <julian.foad_at_wandisco.com> wrote:
> > I'm going to apply this patch or something like it, tomorrow. The new
> > names are very helpful in matching up text-base checksums to the
> > corresponding (old, current, new, copied, revert, ...) text base paths.
> >
> > Trawling through all this is also helping me to understand the text-base
> > manipulations in the update editor, and my next move here will be to
> > replace the old text-base paths with new WC-NG pristine store functions,
> > which is now looking pretty straightforward in this part of the code.
> > I'll just do any stream wrapping necessary to ensure the MD5 checksums
> > are still calculated as well as the new SHA1s.
> >
> > Any comments about this renaming? A mass rename could affect anybody's
> > patches in progress but it should be simple enough to merge. Let me
> > know if there's a particular reason to hold off for a short time.
> >
> > - Julian
> >
> >
Received on 2010-04-08 00:27:44 CEST