2009/2/4 Paweł Hajdan Jr. <phajdan...@chromium.org>:
> I recently tried to port chrome/browser/safe_browsing/protocol_manager.cc to
> Linux, but there's one problem. ProtocolManager uses hash_map of
> URLFetcher*-s, and GCC needs a hash function for it.
>
> I see many ways we can deal with that, and I discussed some of them on irc
> and in review. The result so far is that it would be better to discuss this
> further on chromium-dev. So, what we could do:
>
> 1) Just implement needed hash function for gcc in URLFetcher's header file.
> Works, but then it has to be done in every place where it's needed, and we
> have at least one case of that, for example profile.h.
> 2) Provide default identity hash function for object pointers (it's probably
> what MSVC does on Windows). You can see a patch here:
> http://codereview.chromium.org/20041 . There are concerns about making this
> a default behavior, so...
> 3) Maybe create a macro like GCC_IDENTITY_HASH(ClassName) so we don't have
> to repeat the implementation of identity hash function. Then it's simple to
> apply it explicitly only where it's needed. But well, it's a macro.
>
> Generally each of these choices have some drawbacks, but probably they're
> not the only ways to deal with this problem.
>
> So, if you have any ideas, comments, etc. - please share.
I am not familiar with this API, but in general, I think we should
just use std::map instead. Maybe a safe browsing expert can comment
about this. If we have less than a few hundred or so it doesn't
matter.
Brett
--~--~---------~--~----~------------~-------~--~----~
Chromium Developers mailing list: chromium-dev@googlegroups.com
View archives, change email options, or unsubscribe:
http://groups.google.com/group/chromium-dev
-~----------~----~----~----~------~----~------~--~---
