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

Re: svn commit: r11285 - trunk/subversion/mod_dav_svn

From: <kfogel_at_collab.net>
Date: 2004-10-13 21:45:20 CEST

Tobias Ringström <tobias@ringstrom.mine.nu> writes:
> Thanks a lot for the review and the helpful comments, Karl. I've been
> wrestling with the docstring quite a bit now, and I've come up with
> the following proposal which is quite different from the original
> one. What do you think?

Much better! Comments below.

> /* Set *PROPVAL to the value for the revision property PROPNAME on
> COMMITTED_REV, in the repository identified by RESOURCE, if
> RESOURCE's path is readable. If it is not readable, SVN_NO_ERROR
> is returned and *PROPVAL is set to NULL. Use POOL for temporary
> allocations and the allocation of *PROPVAL.

That paragraph is great. But don't switch to the passive voice for
the second sentence. Just say "If it is not readable, set *PROPVAL to
NULL and return SVN_NO_ERROR."

> Note that this function does not check the readability of the
> revision property, but the readability of a path. The true
> readability of a revision property is determined by investigating
> the readability of all changed paths in the revision. For certain
> revision properties (e.g. svn:author and svn:date) to be readable,
> it is enough if at least one changed path is readable. When we
> already have a changed path, we can skip the check for the other
> changed paths in the revision and save a lot of work. This means
> that we will make a mistake when our path is unreadable and another
> changed path is readable, but we will at least only hide too much
> and not leak any protected properties.
>
> WARNING: This method of only checking the readability of a path is
> only valid to get revision properties for which it is enough if at
> least one changed path is readable. Using this function to get
> revision properties for which all changed paths must be readable
> might leak protected information because we will only test the
> readability of a single changed path.
> */

Nice warning. So, question: should this function just error if the
requested property is one which requires *all* changed paths to be
readable?

-Karl

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Oct 13 23:34:55 2004

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