[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

copy-on-update: question for ehu (or others)

From: Ben Collins-Sussman <sussman_at_red-bean.com>
Date: 2007-10-11 20:59:27 CEST

Erik, since you're a working copy expert at this point, I need you
tell me if my new copy-on-update code has created a bug or not. I'm
not quite sure!

Prior to my changes, the update_editor would only receive simple
add_file() calls with no copyfrom args. The entire file would be
sent by the server: all text, all props (including about 5
"entryprops").

In the new system, add_file() sometimes receives copyfrom args, and is
sometimes able to make use of them. If it's able find the
already-existing file in the wc, then it does this algorithm:

  - copy the text-base over (so merge_file() can install it)
  - push the base-props into the file_baton (so merge_file() can
    install them)
  - run merge_file()
  - if there are text-edits, copy working-file over
  - if there are prop-edits, push them via svn_wc_prop_set().

The result seems to be good, but notice what's missing: the
entry-props! We never asked for any data from the server, so we don't
have any of the things like 'last-changed-author',
'last-committed-date', and so on. When I look at .svn/entries, that
sort of stuff just isn't there after my tests run.

Yet, when I run 'svn info', the entryprops *do* show up. Is there
magical inheritance going on from the parent? (In which case, the
ones being displayed by 'svn info' are just wrong?) Or are the
entryprops that live in .svn/entries truly a cache, and if not
present, fetched from the server?

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Oct 11 20:59:38 2007

This is an archived mail posted to the Subversion Dev mailing list.