gEDA-user: 2 make errors installing gwave

[Werner Hoch]

Hi Dan and all,

On Samstag, 1. Dezember 2007, Dan McMahill wrote:
> Werner Hoch wrote:
> > Yes, but I'd still would like to have a "standard" binary format.
> > hdf5 would be nice.
> > http://hdf.ncsa.uiuc.edu/products/hdf5/index.html
>
> How well to ascii output files scale when you want to write out
> 30,000 node voltages and be able to pick out one to plot without it
> taking a long time?  I don't know the answer, but it seems like a
> binary format could have advantages there.
>
> Like Werner, I've also wondered about hdf5 since there is a library
> available for it and octave and I believe scilab both support it.  

I guess nearly every math package supports hdf5. A quick google search 
tells that you have support for R, ruby, python, perl, ...

Appended is a short python script that writes a table using the pytables 
modul. It takes 10 seconds to write a table with 1 million entries and 
creates a file with ~60MB.

> My 
> weak understanding of hdf5 is that it would meet some of the
> requirements for a simulator output file (you can write many
> variables a sample at a time instead of .mat file where you write all
> the samples of 1 variable followed by all the samples of another). 

That's right.

> It also appears that it would not nearly be enough to just say "hdf5"
> but you'd have to further specify the file format.

Yes, but it's quit easy doing it in the hierarchical structur of hdf5:

simulation_n --> plot_n --> table
or 
simulation_n --> plot_n --> metadata
simulation_n --> plot_n --> vector_n

> In terms of commercial simulators, it is annoying as all get out that
> they insist on wanting to use their own proprietary binary formats
> for results because sometimes you really want to get the data into a
> different tool.  grrrrrr.  This is clearly an area where geda can
> hands down win just by documenting the file format used.

Fully agreed.

Regards
Werner
-------------- next part --------------
A non-text attachment was scrubbed...
Name: hdf5test2.py
Type: text/x-python
Size: 1385 bytes
Desc: not available
Url : http://www.seul.org/pipermail/geda-user/attachments/20071202/00843afb/attachment.py