Re: Final call for bug reports

[Daniel J Sebald]

On 10/16/2013 11:34 AM, Torsten wrote:
> On 16.10.2013 08:40, Daniel J Sebald wrote:
> > 1) I see this at startup quite often.
> >
> > HDF5: infinite loop closing library
> >
> > D,T,F,FD,P,FD,P,FD,P,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E,E
> >
> >
> > I'm not quite sure how to reproduce the issue, however.
>
> Really at startup? I always see this on exit (see bug #37713).

Just today I started seeing this at exit.  The reason I might have been 
seeing it at startup is because of plot/graphics errors...maybe.  I see 
the link to a discussion of this bug in Savanah.  In the following test 
file is also a little more detail:

http://www.hdfgroup.org/ftp/HDF5/current/src/unpacked/hl/test/test_file_image.c

It has something to do with a memory reallocation failing, at least in 
that description.  But actually, the codes are different so Octave may 
have a different problem.

I quickly looked at the HD5 source code that is generating this sequence 
of codes.  (It's in H5.c.)  It is designed to shut down the various 
interfaces that (possibly mutliple?) APIs are using.  It first goes 
through the higher level interfaces and if those shutdown successfully, 
proceed to shutdown the lower level interfaces that the higher level 
interfaces might need.  If the close wasn't successful, the loop breaks 
and tries again, up to 100 times.  Basically, the shutdown keeps looping 
through the higher level interfaces until it can move on.

So, if we look at the above code sequence, it is closing a few high 
level interfaces without any problems.  Then, there are a some midlevel 
interfaces, FD and P which take a few passes to shutdown.  Once those 
are successful, the routine moves on to the Error Code interface, E. 
Clearly, HD5 is failing to close the Error Code interface.  Why, I don't 
know.

Can anyone think of whether Octave might use/not use the HD5 error code 
in recent versions of packages that didn't happen in the past.  (I think 
this started for me when I began using a few packages.)


> > 3) What are the default font settings?  My preference is a mono-space
> > font, so I've chosen Liberation Mono 11 point as the Default and all
> > other text types to have the Default font.  I felt that 9 point is kind
> > of small for editing.  9 point suffices for the Octave output.  When the
> > comments are not the same font as the rest of the code and nothing is
> > mono-spaced, layout didn't look very organized.
>
> The default font settings are those from the qscintilla lexers. The
> default font names depend on the operating system. It would be easy to
> supply other settings with the default settings file but then we would
> have to take care about the existence of fonts on different platforms.

There must be a set of fonts that goes across platforms.  Or perhaps a 
routine for a suggested font based upon a more general description?  I 
would look into this more, but time is limited.

Dan