RE: [Axiom-developer] RE: gcl-2.6.8pre on MAC OSX 10.2

[Bill Page]

On November 4, 2006 12:42 AM Gaby wrote:
>...
> Bill Page wrote: 
> | I think the following two files are not used anywhere:
> | 
> | aTx=b.bitmap
> | ATX=B.bitmap
> | 
> | (See list of unused bitmap files below.)
> 
> How do we determine that they are referenced at run-time through
> dynamic generation of their names? It is because we know no such
> things exist?

I do not know that they are not referenced at run-time through
dynamic generation of their names, however given what I know about
this programs it seems very unlikely to me.

> 
> | > Also in 'src/doc/ps' we have 'SEGBIND.ps' and 'segbind.ps'.  
> | > 'segbind.ps' is a screen dump of a graphic window containing a
> | > parabola.  I can not really see 'SEGBIND.ps' because it triggers
> | > Postscript stack error in ghostscript (like many other files in
> | > 'src/doc/ps' :(.
> | > 
> | 
> | I think all .ps files that are not referenced in any *.pamphlet file
> | (such as SEGBIND.ps and segbind.ps) should be deleted from the /ps
> | directory.
> 
> Before deletion, we have to determine that they are not referenced
> at runtime through name generation on the fly.  Do you know whether
> that can ever happen?
> 

I do not know and I think it might be prohibitive in time to examine
the code in sufficient detail to ensure that this is not the case but
it seems to me the benefits of cleaning up this mess out weight the
risk, especially since these files are among the problematic binary
and mixed-case files in the distribution.

Here is one suggestion: For now we could just move all of the "unused"
files that I listed in my previous email to temporary location in the
source tree. The recent improvements to hyperdoc made by Waldek suggest
that now would be a good time in any case to have Axiom users focus
more attention on this part of the system. If problems of missing
files occur, it should be difficult to document which of the presumed
"unused" files is actually referenced dynamically (if any).

Regards,
Bill Page.