Front page | perl.perl5.porters |
Postings from October 2003
On Fri, Oct 03, 2003 at 08:01:05AM +0200, "H.Merijn Brand" <h.m.brand@hccnet.nl> wrote: > On Thu 02 Oct 2003 22:46, Andy Dougherty <doughera@lafayette.edu> wrote: > > On Thu, 2 Oct 2003 schwern@pobox.com wrote: > > > > > Yes, some sort of "just use the biggest numbers you can" option would be > > > nice. > > > > Yes, it would be nice, but it's *very hard* to actually implement it well. > > Why? > > make -Dusemanybits be equivalent to -Duse64bitall -Duselongdouble, but set a > flag that if any of these would fail to just continue. Which is more or less > regaining the situation from before these (IMHO correct) changes to Configure > with the exception that it is now a *chosen* option to continue, and that some > hints files need some twiddling to not die in these occasions. > > Let me make clear that the user that would choose this option will not have > any guarantee that perls will be configured the same accross OS's, which was > one of the reasons to make these patches in the first place. If the user chose > 64bitall, he was not guaranteed to get it in the old situation, making > portability accross systems harder. > > OTOH upgrading is still allowed (and done). i.e. on HP-UX choosing > -Duse64bitint will upgrade to -Duse64bitall The problem AIUI is that for part of the Configure run, use64bitall is set, and the hints (or even the units) may make choices based on that. And then when it is turned off, those choices go awry. The easiest solution is probably to restart Configure with modified arguments.Thread Previous | Thread Next