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

Re: ra_svn vwrite_tuple() optional elements robustness

From: Stefan Fuhrmann <stefan2_at_apache.org>
Date: Sun, 21 May 2017 20:09:11 +0200

On 15.11.2016 00:43, Daniel Shahaf wrote:
> Stefan Fuhrmann wrote on Mon, Nov 14, 2016 at 12:50:51 +0100:
>> On 06.11.2016 02:21, Daniel Shahaf wrote:
>>> During the r1767197 thread, I noticed that vwrite_tuple() doesn't
>>> enforce its precondition that in the optional part of a tuple, either
>>> all values are valid or no value is; a call such as
>>>
>>> vwrite_tuple("(?r?r)", SVN_INVALID_REVNUM, (svn_revnum_t) 42)
>>>
>>> would happily generate a «( 42 ) » tuple, instead of triggering an
>>> SVN_ERR_MALFUNCTION().
>>>
>>> This would be easy to fix, and would prevent a class of bugs.
>> This function seems to have a number of bugs,
>> for instance:
>>
>> * optional sub-structs don't get a "("
>> * the OPT flag is global instead of per-struct
>
> I hope at least the parser routines don't have such bugs.
>
> (excuse brevity; can't be verbose at the moment)

As it turns out, the current implementation is correct but
forbids optional sub-tuples. See r1795714.

-- Stefan^2.
Received on 2017-05-21 20:09:22 CEST

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