Karl Fogel <kfogel@newton.ch.collab.net> writes:
> cmpilato@collab.net writes:
> > Heh, I wonder if this means that the modify_entry function will now
> > look something like:
> >
> > modify_entry (svn_wc_entry_t *entry, /* the entry to modify */
> > svn_wc_entry_t *new_vals, /* a nice place to hold new values */
> > apr_uint16_t modify_flags, /* bit-mask of which values to mod */
> > apr_pool_t *pool) /* our favoritest structure in svn */
>
> Nah, I think we just wouldn't *have* a modify_entry() function.
> Instead, you just, you know, modify the entry, using the C assignment
> operators. The read/write functions do type conversion for you (since
> everything in a .entries file is a string, but the structure wants to
> hold svn_revnum_t, etc).
Well, careful now. modify_entry() exist so that changes to an entry
can be "merged" in before being written to disk. That is, if I have
an entry scheduled for addition, and I "merge" in a deletion, the
entry goes away altogether. Or if it's deleted, and I merge in an
addition, it becomes "replaced". That kinda thing. We don't want
that logic spread all over the place.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Mar 14 16:56:22 2002