[Lldb-commits] [lldb] r203107 - Fix Windows build break introduced in r203035.

[Virgile Bello]

Yes it would end up being mostly SVN move (differences are very small,
mostly sizeof(GPR::register) macros become sizeof(((GPR*)NULL)->register)
so that it works with MSVC.
In that case I suppose I better wait for Todd's changes? (sounds great BTW!)

Note that Windows build is currently broken, so please let me know if you
think it will take some time to get those changes sorted out, I will just
put a temporary #ifndef _WIN32 around those two modules init in lldb.cpp in
that case.



On 11 March 2014 06:02, Todd Fiala <tfiala at google.com> wrote:

> As an early heads up on this subject: for lldb-gdbserver, I'm in the early
> processes of refactoring ProcessPOSIX and derived classes to decouple
> ProcessMonitor from ProcessLinux/ProcessFreeBSD.  The way I'm doing it,
> ProcessMonitor is essentially going to become embedded in a
> NativeProcessProtocol.  Just like now, I expect there will be a separate
> one for both FreeBSD and Linux.
>
> For the case of lldb-gdbserver where we don't want a heavy-weight
> ProcessPOSIX-like object running the show, the same ProcessMonitor kernel
> of operation (via the NativeProcessProtocol impls, communicating via the
> Broadcaster/Listener mechanism) will be used.
>
> I mention this because I think the decoupling that will be introduced by
> this whole process will likely shake out some of the duplicate code that
> exists on the Linux and FreeBSD side (and/or identify code that is
> definitely not platform neutral).  So some of what you mention here:
>
> > As an aside, note that the division of code between the POSIX and
> FreeBSD/Linux classes isn't quite correct; in some cases identical or
> nearly-identical code exists in both the FreeBSD and Linux classes,
> and in other cases code in ProcessPOSIX is actually Linux specific.
>
> might just shake out or become more approachable after I pound through my
> changes.
>
>
> On Mon, Mar 10, 2014 at 1:44 PM, Ed Maste <emaste at freebsd.org> wrote:
>
>>  Seems fine to me as well, assuming it's actually "svn mv" when you go
>> to commit.  If it compiles on Linux it should on FreeBSD as well, or
>> will be trivially fixed.
>>
>> As an aside, note that the division of code between the POSIX and
>> FreeBSD/Linux classes isn't quite correct; in some cases identical or
>> nearly-identical code exists in both the FreeBSD and Linux classes,
>> and in other cases code in ProcessPOSIX is actually Linux specific.
>> It might be a good opportunity for us to clean some of this up if
>> you're going to introduce base classes for Windows and POSIX.
>>
>> On 10 March 2014 14:12, Greg Clayton <gclayton at apple.com> wrote:
>> > Looks fine.
>> >
>> > On Mar 10, 2014, at 9:16 AM, Virgile Bello <virgile.bello at gmail.com>
>> wrote:
>> >
>> >> Small update: looks like it compiles fine on Linux.
>> >> Let me know if I should commit.
>> _______________________________________________
>> lldb-commits mailing list
>> lldb-commits at cs.uiuc.edu
>> http://lists.cs.uiuc.edu/mailman/listinfo/lldb-commits
>>
>
>
>
> --
>  Todd Fiala | Software Engineer |  tfiala at google.com |  650-943-3180
>
> _______________________________________________
> lldb-commits mailing list
> lldb-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/lldb-commits
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/lldb-commits/attachments/20140311/92edd5d1/attachment.html>