[Omake] Bug in Lm_db w.r.t. overwriting existing entries?

[Aleksey Nogin]

On 29.05.2007 17:48, Aleksey Nogin wrote:

> [...] in the bug 669 example, If I run 
> the commands below, all those marked with "<--" would evaluate a static 
> section. Those that are marked "???" do not have a valid reason for 
> doing so.
> 
> rm -f common/*omc
> (cd a; omake)      #<--
> (cd b; omake)      #<--
> (cd b; omake)      #<-- ???
> (cd b; omake)
> (cd a; omake)      #<-- ???
> (cd a; omake)      #<-- ???
> (cd a; omake)
> 
> Could it be that the ref cell hack just moves the problem somewhere else?
> 
As far as I can tell, the "ref cell hack" is working completely 
correctly, but we've just uncovered a different bug somewhere else - 
probably in Lm_db.

Basically,

% rm -f common/*omc
% (cd a; omake)

Would add a table with "$static1 -> ..." to the .omc under the tag "2" 
("values_tag").

% (cd b; omake)

Would then:
   - load the "$static1 -> ..." table from the .omc
   - add the "$static2 -> ..." entry to the table
   - Lm_db would then claim to write the updated table back to .omc

So far so good. Now I do

% (cd b; omake)

again. Now Lm_db says it's unable to find the requisite entry in the 
.omc! I am not sure whether the save was incorrect (the .omc does grow, 
but that's all I know so far), or whether the bug is in the load code.

Aleksey
-- 
Aleksey Nogin, Research Scientist
Advanced Technologies Department, Information & System Sciences Lab
HRL Laboratories, LLC, Malibu, CA