On 08/06/2016 03:50 PM, Jonas Smedegaard wrote:
> Hi Roberto,
>
> Quoting Roberto Bagnara (2016-08-06 14:34:14)
>> On 08/06/2016 02:27 PM, Jonas Smedegaard wrote:
>>> Quoting Dmitry Tsarkov (2016-08-04 22:09:56)
>>>> I'm the implementer of the original FaCT++ system. After checking
>>>> the project's makefiles I found out that the dependency from the ppl
>>>> package is optional. The user might use additional commands to turn
>>>> on options that require this package, but by default it is not
>>>> needed. Could the dependencies for the Debian package be adjusted to
>>>> reflect that fact? Alternatively I can make an intermediate release
>>>> to completely remove the offending options.
> [...]
>>> If you can tell - e.g. by providing a patch - how to suppress ppl
>>> then that would be nice, and adequate for our redistribution of
>>> FaCT++ in case you prefer for postpone a release till you have other
>>> more exciting changes.
>>
>> May I ask why ppl should be suppressed at all?
>> The new version upstream (PPL 1.2, released in February 2016) solves
>> all problems wrt GCC 6. If upgrading to the latest upstream release
>> is not wanted (why?), then patches have been provided in this very
>> issue. Kind regards,
>
> Whoops. Seems Dmitry (ill-adviced by me) posted his comment to a wrong
> bugreport: FaCT++ is linked with a different ppl - not Debian-packaged
> "ppl" from from http://www.cs.unipr.it/ppl/ but "cloog-ppl" from
> http://www.CLooG.org/ .
>
> Sorry everyone for the confusion!
>
> To answer your question, Roberto (if still relevant): I simply trust
> Dmitry as author of FaCT++ when he states that "by default [ppl] is not
> needed".
>
> If ppl is somehow better than cloog-ppl, then perhaps now is a chance
> to try convince Dmitry to switch?
If I understand correctly, cloog-ppl is an obsolete library that
uses (an old version of) ppl (i.e., the Parma Polyhedra Library),
which does not provide the same functionality. Moreover, cloog-ppl
cannot be replaced by ppl alone. Summarizing, my understanding
is that the discussion on FaCT++ has nothing to do with this issue.
Kind regards,
Roberto
--
Prof. Roberto Bagnara
Applied Formal Methods Laboratory - University of Parma, Italy
mailto:[email protected]
BUGSENG srl - http://bugseng.com
mailto:[email protected]
