[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [PATCH 5/5] Generate a x_server_routine in the sheader so it can be
From: |
Justus Winter |
Subject: |
Re: [PATCH 5/5] Generate a x_server_routine in the sheader so it can be inlined |
Date: |
Mon, 02 Dec 2013 17:25:23 +0100 |
User-agent: |
alot/0.3.4 |
Quoting Samuel Thibault (2013-12-02 13:09:44)
> Well, it means that whenever a protocol gets augmented, one has to
> recompile all demuxer functions calling these inlines. In the example
> of trivfs_demuxer, the application itself does not need a rebuild, only
> libtrivfs needs, but in other cases that will not be true. This is a
> quite subtle issue we need to make sure won't happen, by making sure
> all libraries implementing RPCs provide a demuxer function, and don't
> just rely on their caller calling the proper server_routine function.
> Perhaps it's already the case, and then I'm fine with mig generating
> inlines.
Note that for inlining without -flto, the definition has to be known
at compile time. Currently, the demuxers won't inline the functions,
even if mig would include this declaration in the generated x_S.h
file, b/c it is not #included. So for current demuxers this won't
change anything, they will continue to use the one from xServer.c.
Justus
- [PATCH 1/5] Advise flex not to generate the input function, Justus Winter, 2013/12/02
- [PATCH 2/5] Avoid a compiler warning in WriteDefines, Justus Winter, 2013/12/02
- [PATCH 3/5] Clean up generated sources, Justus Winter, 2013/12/02
- [PATCH 4/5] Move the generation of x_server_routine function into a function, Justus Winter, 2013/12/02
- [PATCH 5/5] Generate a x_server_routine in the sheader so it can be inlined, Justus Winter, 2013/12/02
- Re: [PATCH 1/5] Advise flex not to generate the input function, Samuel Thibault, 2013/12/02