Re: [BUILD] multi-platform/OS test needed
"eric.bachard" <[email protected]>
| Newsgroups | gmane.comp.openoffice.devel.porting |
|---|---|
| Message-ID | <[email protected]> |
Hi Eike, Eike Rathke a écrit : > Yes. If math.h isn't needed for anything else in that file, I didn't > check that, but it didn't look like by browsing through. I just found finite() too, if I'm not wrong. > ?? why should the changes be removed? Replacing isnan() by > ::rtl::math::isNan() solves porting issues, as may the replacement of > finite() with ::rtl::math::isFinite() though it seems there weren't any > issues with that by now. It's not very important, I simply thought they have to be done simultaneously, because finite() needs math.h inclusion, I just have deleted. IMHO , if the changes are made separately, we have something like "undefined" waiting the moment the two changes are effectives. But I probably am wrong... To be sure, I propose to complete the second part, changing both isnan() in ::rtl::math::isNan() and finite() in ::rtl::math::isFinite(). If you agree, these changes will be in macosx07 cws, and you or Maho can complete in another, deleting the FREEBSD part. Just for make logical changes. Issue 49079 is about replacing the entire > string/double conversion code with calls to analogous functions in > ::rtl::math, but that is not relevant for porting right now, it only > would remove duplicated code and assure that if there were any porting > issues in future they could be fixed in one central place. Yes, for this I have seen, this part is exactly #i49079#. Regards, eric bachard -- eric bachard <[email protected]> Francophone OpenOffice.org Commmunity developer (Linux PPC / Mac OS X / X11) See : <http://fr.openoffice.org>