Investigating bug 4053

[François Vogel]

Hi,

I'm currently investigating bug 4053:
http://bugzilla.scilab.org/show_bug.cgi?id=4053
and this one is quite puzzling.


After a fair amount of investigations in different directions, I think 
that maybe part of the problem is in this commit
3aae8bc45b1a37eadcd495655cb85104881c8a38
from Allan Cornet on 26 August 2008. At least it is suspicious to me 
for the reasons I will try to explain.

This commit just adds a test on the existence of the scipad Tcl 
interpreter in somewhat buried Scipad ancillary. The commit comment 
does not tell why this test was deemed necessary.

I would make two comments on this commit:

1. This test should be useless because it should always return true: 
proc setscipadwords is not supposed to be called by something else 
than Scipad, thus the scipad Tcl interp always exists when this proc 
gets called.

2. There is no "else" clause, and this would lead to empty keywords 
arrays for colorization.


These observations in turn bring two questions:

a. Why did you think at some point of time that this test was needed? 
Can you exhibit a case where it returns %F?

b. If this test has good reasons to be there because it may return %F 
in some case, then what do you think should happen in the missing else 
clause?


Thanks,
Francois