Re: svn commit: r983807 - in /subversion/branches/atomic-revprop/subversion: include/svn_error.h libsvn_subr/error.c
From: Daniel Shahaf <d.s_at_daniel.shahaf.name>
Date: Tue, 10 Aug 2010 15:04:07 +0300
(I intended to commit that to trunk)
Julian Foad wrote on Tue, Aug 10, 2010 at 12:17:24 +0100:
I'll need it to dig beneath ra_svn's wrapping of server-generated errors by
> I would expect to be able to call such a function on a NULL error
Ahem, doesn't (apr_status_t)APR_SUCCESS mean "no error"?
To make svn_error_has_cause(SVN_NO_ERROR, apr_err) return TRUE if and only
> The behaviour I mentioned above would fall out
I didn't consider the option that APR_SUCCESS could occur as part of the
> - Julian
I can see several options:
* forbid passing SVN_NO_ERROR
Right now, the API does the first and the code the third.
I suppose the question is (1) whether we expect the caller to test for
This is an archived mail posted to the Subversion Dev mailing list.