[Simh] Readline support gone?
Philipp Hachtmann
hachti at hachti.de
Wed Sep 17 10:12:46 EDT 2008
BTW, the list's reply settings are a bit..... unsophisticated *g*
>> A set of real Makefiles would be nice. As a project grows, the
>> importance of a makefile becomes even more evident. Since gmake is
>> available for many platforms, a single makefile for each simulator
>> should work, even when using the Microsoft compiler. And there could
>> be a top-level makefile that descends into each similator subdirectory
>> and invokes a simulator-specific makefile.
Yes.
>
> Bearing in mind RMCH.
I just read the article. It mentiones some effects to keep in mind. The
best performance would be achieved with a single makefile, indeed. But
traversing those few directories isn't a high price for a lot more clarity.
>>
>> I guess it would take some time to debug the makefile on all supported
>> platforms, but it would be worth it in my opinion.
>
> Sounds like Philipp already did a lot of it.
I once made some changes. This was a long time ago.
Here you can find an archive containing my branch:
http://hachti.de/download/simhv33-1-hachti.tar.bz2
You see, it's not the last version.
- all directories in lower case
- no sources in top directory -> moved to sim folder
- depends on gmake
- original "makefile" is still included
- readline support with history - but no tab completion of simulator
commands.
Keeping in mind the RMCH article, the best performance would be achieved
with a single real makefile for the whole project. That would be a great
opportunity.
It's not ok to put all options into one CC command. It's way better to
compile to objects and link, using CFLAGS and LDFLAGS.
With a real makefile you can take full advantage of multiprocessor
systems using the -j switch of make. On my quad core that improves the
linux build time by more than factor 3.5!
Best wishes,
Philipp :-)
--
http://www.hachti.de
More information about the Simh
mailing list