Danek Duvall wrote:
> On Wed, Mar 11, 2009 at 09:41:25AM -0700, George Vasick wrote:
> 
>>> Is there any reason that 1.10 can't simply be replaced by 1.10.2?  IIRC,
>>> there were incompatibilities between 1.9 and 1.10, which is why both are
>>> there.  As far as I can tell, Ubuntu doesn't have separate packages for
>>> each version in the 1.10.X series, just for those in 1.X.
>> Yes, I think this is a better approach.  I'll do some test builds with SFW 
>> and ON to make sure they are compatible.
> 
> Cool, thanks.
> 
>>> Also, note that most of the updates you present here have no interface
>>> changes, so they don't really need ARC approval -- simply upgrading one
>>> component to another doesn't necessarily have any architectural impact.
>> What about adding a man page where the previous version of a package did 
>> not include one or adding new info pages that were not in the previous 
>> version?
>>
>> Technically, this could be viewed as a change to the user interface, but
>> it wouldn't necessarily have architectural impact.
> 
> Documentation may be a part of the user's total experience, but it isn't an
> interface, so it doesn't have architectural impact.

Hi Danek,

I reviewed changes between the 1.4.2 and 1.4.12 version of the GNU m4 
preprocessor.  I found two new options:

--debugfile=FILE
--warn-macro-sequence[=REGEXP]

These add additional diagnostic capability compared to the earlier release.

I also ran the 1.4.2 test suite against version 1.4.12.  I found four 
differences in the results:

- 3 due to formatting changes in in error diagnostics.  The messages are 
no longer prefixed with "m4:".
- 1 due to a change in the WIDTH operand to eval().  Leading minus signs 
are now excluded when calculating the width of a field.

Would this level of change be appropriate for an ARC review?


Thanks,
George


> 
> Danek

Reply via email to