Hello,

> For the last couple of months I've been toying with the notion
> of an [HDF5][1] based signal file for GHDL.
> 
> [1]: http://tinyurl.com/pafyvj7

Last time I have looked at HDF5, the copyright of HDF5 writers
don't look to be GPL compliant.  Is that correct or am I wrong ?

Generating HDF5 from ghdl shouldn't be very complex - just mimic
the vcd or ghw writer.

> GTKwave does not read HDF5 files (but many other data analysis
> tools do). There is, of course, a C language API to the HDF5
> libraries but [Ada language bindings][2] haven't really
> materialized yet.
> 
> [2]: http://tinyurl.com/cawqwdh

Interfacing C and Ada isn't hard at all (at least for simple types).
I am not worried by that at all!

> Needless to say, extending GHDL and GTKwave with HDF5 support
> would be a significant engineering project but the possibility
> of having richly-typed, self-describing data structures in a
> portable, high-performance file format is intriguing.

I suppose HDF5 is adapted for anagogic signals, but I also suppose
not well for digital one, i.e.: compression isn't good compared
to other format.  But I haven't looked.

Maybe HDF5 is as good as GHW!

Tristan.

_______________________________________________
Ghdl-discuss mailing list
[email protected]
https://mail.gna.org/listinfo/ghdl-discuss

Reply via email to