On Tue, 2009-07-21 at 15:51 +0200, Pierre Marchand wrote: > Vous (Pierre Marchand) avez écrit : > > I’m getting into your recent changes and had a couple of trivial issues I > > wan’t to know if I must fix it myself or wait for you finish the work. > > > > + podofo_config.h is not installed. > > + HAVE_FONTCONFIG is publicly defined in podofo_config.h but not prefixed > > with "PODOFO_". > > Going further I see there are rather more changes than I thought at first > sight. > It’s still not clear for me if podofo_config.h is intended to be public or > just used at build time, I did notice it was missing when building apps > relying on podofo that complained PDF_INT*_TYPENAME were not naming types.
Hmm, that's a really good point, in that I've failed to separate what should be public and what should be private (podofo compile-time only). *headdesk* After all the work I did getting Pdf3rdPtyForwardDecl.h etc set up to hide PoDoFo's external header dependencies you'd think I'd not do somthing this dumb. Sorry. I do think moving to the config.h approach will be clearer, I just need to split it (and the compiler compat stuff) into public/private parts and install the public parts. I'll see if I can look at it tomorrow. > Regarding deprecation of pdf_long as a ptrdiff_t, the reasons for it are > quite clear I'd _really_ appreciate an explanation, then, because I'm struggling with understanding the rationale behind how is use as a typedef for ptrdiff_t makes sense in some places. In most of the places I see it used, size_t would be more appropriate, and direct use of ptrdiff_t would be clearer in most of the others. It's also used in things like handling of the PDF xref table, where handling of >2GB files is desirable. That'll only work on a 64-bit machine with pdf_long as a ptrdiff_t . I think "pdf_long" is too generic a name and lends its self to use for inappropriate things. If you want a string or buffer size, use size_t. If you want an offset, use ptrdiff_t directly. For file/stream work, I'm inclined to suggest using the STL typedefs: If you want a file offset use std::streamoff and for a file position std::streampos . "pdf_long" is meaningless and generic, and IMO rather confusing as it obscures both purpose and size/signedness. That said, I was certainly wrong to change it in the sources as I did. My only defense is that I was really tired and too thick to notice it. I've reverted the alteration of the typedef. I do think pdf_long needs to disappear though, and that it's currently used incorrectly where it touches on file offsets, xref offsets, etc. -- Craig Ringer ------------------------------------------------------------------------------ _______________________________________________ Podofo-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/podofo-users
