El dimecres, 28 de febrer de 2018, a les 11:02:11 CET, Oliver Sander va escriure: > On 21.02.2018 00:31, Albert Astals Cid wrote: > > I'd really appreciate other people's comments on this. > > I am generally in favour of changes like this, provided there is no > significant increase in run-time or binary code size. I think such changes > do decrease long-term maintenance costs, and they make it easier for new > developers to understand the code. The current GooHash implementation may > be a reasonable implementation of a hash table (I didn't check), but it is > completely undocumented, and if I ever need to know what it does, I will > have to read the code. On the other hand, while that emplaceRangeMap lambda > looks strange to me on first sight too, all this modern C++ stuff is well > described in books, online articles and blog posts. > BTW: I always thought that changes like this couldn't be made because it > makes backporting patches from xpdf more difficult/impossible. Is this not > an issue anymore?
we never got around to merge xpdf 3.04 changes and given the relatively big change i did regarding Object i guess we can say that porting xpdf code isn't really a big motivation anymore for not making improving changes. Cheers, Albert > > Best, > Oliver _______________________________________________ poppler mailing list poppler@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/poppler