On 14-10-2009, Stefano Zacchiroli <[email protected]> wrote: > > --C7zPtVaVf+AK4Oqc > Content-Type: text/plain; charset=iso-8859-1 > Content-Disposition: inline > Content-Transfer-Encoding: quoted-printable > > On Wed, Oct 14, 2009 at 09:40:49AM +0000, Sylvain Le Gall wrote: >> >> If toto.cma has been compiled with -g, this is at least a warning (I >> >> would even say it is an error). >> > >> > May I remind you that our current policy recommends ("should") compiling >> > all objects with debugging enabled? Chapter 3, on line version at >> > http://pkg-ocaml-maint.alioth.debian.org/ocaml_packaging_policy.html/c3= > 05.h=3D >> > tml >>=20 >> I think it deserve a removal from the policy ;-)=20 >> Do you agree ? > > No, not really, but let's proceed in stages. > > I concur that thus far we haven't done a good job of enforcing that > policy recommendation (the fact you, which are very active in d-o-m, > were unaware of it is already a good evidence of that). Then, even if I > don't remember exactly the history, I'm confident that the > recommendation was not added by a "lone wolf" :-), we reached a more or > less rough consensus before somebody added it to policy. Before > reharshing the topic, it would be nice if someone can dig up the history > of that addition. > > One argument for not removing the "use -g recommendation" is the one > given by St=E9phane: -g has no runtime penalty. Also, I add that it is > very useful and badly needed by programmers: when your program fails, > you want to understand _where_, no matter if it is in your code or in a > system library. When the failure is in a system library, it does not > necessarily mean it is a but in that library, it might have been you > that violated some assumption described in the lib API. > > That latter aspect is what, in my mind, makes "-g" different from > "-p". When I do profiling I'm usually interested in profiling my own > code, not that of the lib I'm using. Surely sometime you need to do that > as well, but in those cases recompiling them is probably the right way > to proceed, as it is rare enough. > > To conclude: I'm for keeping the recommendation.
I agree with you and Stephane argument is good. Just forget about it. Keep the "-g policy". However the "-p" lintian check is maybe worth. Regards, Sylvain Le Gall -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
