On 29/10/14 22:36, Brian Drummond wrote:
So ... 0.32 has a good subset of VHDL-2008 support.
I have implemented features for which I have found an existing
needs in some units. If anyone has a vhdl-2008 design that isn't
supported by ghdl, feel free to send it!
I know you tried OSVVM 2014.1, and though 2014.02 was talked about, I
believe 2014.1 is still current. That should please Jim Lewis!
1) Does it support Daniel Kho's VHDL-2008 feature requests (on gna.org)?
Yes for process (all), no for block comments.
2) What is the intended status of the backends? (subject to testing in
this phase...
a) gcc of course
Yes.
b) mcode (for x86-32) presumably
Yes (that's the one I use daily).
c) is the LLVM backend still experimental? Should we announce it and
start to get feedback from early adopters, with a view to making it part
of 0.33?
Let's say it is still experimental. I don't think we want to support
too many back-ends. We have to decide if we want to release with LLVM
instead of mcode or/and gcc.
d) You have added a "simulator" mode ... what is its status?
I haven't yet updated it for generic packages. Let's forget it for
the release. Performances are very poor (but debugging experience
could be excellent).
Volunteers are welcome for testing or creating releases.
Time I got busy then!
You're welcome (and more than welcome) !
I've been reading your commits ... basically just watching in awe!
Thank you.
For 0.33, I'd like to concentrate on usuability: add
a vzt dumper, improve speed. Adding new vhdl 2008 won't
be the priority for 0.33
OK. Is it possible to look at an efficient "packed array of boolean"
implementation of bit_vector? It might please YG, and set the cat among
the pigeons in the current integer/logical/shift/modular debate if ghdl
was the industry leader in speed there!
I have thought about it a little bit. I think this is doable, although
it may collide with other optimization ideas. But we also need to
do something for std_logic_array, as this is more widely used.
Tristan.
_______________________________________________
Ghdl-discuss mailing list
[email protected]
https://mail.gna.org/listinfo/ghdl-discuss
