Re: testpregap

[Peter Creath]

On Sun, Jun 8, 2008 at 10:04 PM, Robert William Fuller
<hydrologiccycle@...> wrote:

> Robert William Fuller wrote:
>>> Incidentally, testpregap is currently failing the regression tests on
>>> my machine.  Is this a known bug, or is it particular to my setup?
>>
>> Uhhh no it's not a known bug.  Not sure what to do about it....  I don't
>> have MacOS.
>
> By the way, is it a PowerPC or an Intel box?  This should not be a problem
> because I was careful to use the appropriate LE, BE (little endian/big
> endian macros,) but I might have missed something.

It's PowerPC.  That sounds like a likely source of the problem.  I'll
try to chase this bug down.

[Peter Creath]

I just commited a fix for this bug.  It was a structure packing issue.

PowerPC likes to 4-byte align (and pad) structs, so _dao_common_t was
getting expanded to 24 bytes instead of 22.  A carefully placed
GNUC_PACKED directive caused the test to pass.

While I was at it, I fixed some compiler warnings in testpregap.c.in.

[Robert William Fuller]

Peter Creath wrote:
> I just commited a fix for this bug.  It was a structure packing issue.

I did a cvs diff on nrg.h and it claimed no changes!

> PowerPC likes to 4-byte align (and pad) structs, so _dao_common_t was
> getting expanded to 24 bytes instead of 22.  A carefully placed
> GNUC_PACKED directive caused the test to pass.
>
> While I was at it, I fixed some compiler warnings in testpregap.c.in.

Hmm ditto with this....  I wonder if something is wrong with cvs?

[R. Bernstein]

Robert William Fuller writes:
 > Peter Creath wrote:
 > > I just commited a fix for this bug.  It was a structure packing issue.
 >
 > I did a cvs diff on nrg.h and it claimed no changes!

http://cvs.savannah.gnu.org/viewvc/libcdio/libcdio/lib/driver/image/nrg.h?hideattic=0&r1=1.6&r2=1.7&sortby=date

 >
 > > PowerPC likes to 4-byte align (and pad) structs, so _dao_common_t was
 > > getting expanded to 24 bytes instead of 22.  A carefully placed
 > > GNUC_PACKED directive caused the test to pass.
 > >
 > > While I was at it, I fixed some compiler warnings in testpregap.c.in.
 >
 > Hmm ditto with this....  

http://cvs.savannah.gnu.org/viewvc/libcdio/libcdio/test/testpregap.c.in?r1=1.1&r2=1.2&sortby=date

[Robert William Fuller]

R. Bernstein wrote:

> Robert William Fuller writes:
>  > Peter Creath wrote:
>  > > I just commited a fix for this bug.  It was a structure packing issue.
>  >
>  > I did a cvs diff on nrg.h and it claimed no changes!
>
> http://cvs.savannah.gnu.org/viewvc/libcdio/libcdio/lib/driver/image/nrg.h?hideattic=0&r1=1.6&r2=1.7&sortby=date
>
>  >
>  > > PowerPC likes to 4-byte align (and pad) structs, so _dao_common_t was
>  > > getting expanded to 24 bytes instead of 22.  A carefully placed
>  > > GNUC_PACKED directive caused the test to pass.
>  > >
>  > > While I was at it, I fixed some compiler warnings in testpregap.c.in.
>  >
>  > Hmm ditto with this....  
>
> http://cvs.savannah.gnu.org/viewvc/libcdio/libcdio/test/testpregap.c.in?r1=1.1&r2=1.2&sortby=date

Thanks for fixing!

I must be using cvs wrong.  I suppose it only shows local changes....

Regards,

Rob

[Peter Creath]

On Tue, Jun 10, 2008 at 12:01 AM, Robert William Fuller
<hydrologiccycle@...> wrote:
> I must be using cvs wrong.  I suppose it only shows local changes....

Yup, it knows which version you checked out and by default shows the
difference between that version in the CVS repo and your local copy.

You can use explicit revision numbers to query two versions in the
repo.  E.g., "-r1.6 -r1.7".  And "cvs log" will let you figure out
which revisions to compare.

And you can use a single explicit revision number to compare a
specific version in the CVS repo against your local copy.