[ofa-general] Bugs in opensm/libvendor

[Hal Rosenstock]

On Thu, Dec 18, 2008 at 10:48 AM, Sasha Khapyorsky <sashak at voltaire.com> wrote:
> On 10:18 Thu 18 Dec     , Hal Rosenstock wrote:
>>
>> > I don't expect any hurt for osmtest (however didn't check this yet).
>>
>> Depends on what you mean by hurt:
>>
>> osmtest/osmtest.c:      req.query_type = OSMV_QUERY_PATH_REC_BY_PORT_GUIDS;
>> osmtest/osmtest.c:      req.query_type = OSMV_QUERY_PATH_REC_BY_GIDS;
>> osmtest/osmtest.c:      req.query_type = OSMV_QUERY_PATH_REC_BY_LIDS;
>> osmtest/osmtest.c:      req.query_type = OSMV_QUERY_PATH_REC_BY_PORT_GUIDS;
>>
>> So in line with your approach, these instances should be changed over
>> to user specified ones.
>
> Look closer how it is used there

I'm well aware of how it's used there.

> - it is transparent to num_paths
> returned. I don't think we need to change something in osmtest.

In terms of breaking osmtest, it doesn't but it is a semantical change
and no longer will all paths always be able to be validated without
such a change. I know I used to find that useful there.

>> Also, IMO saquery should support both compliant and extended queries.
>
> If somebody cares I can accept the patch with '--complaint' option
> implemented.

I think you mean --noncompliant but anyhow I don't like the backward
compatibility/old semantics not being supported (in addition in these
in tree applications) but not sure I have the time.

-- Hal

> Sasha