Re: MPNumber struct instead of int[MP_SIZE]

[Klaus Niederkrueger umpa ens-lyon fr]

> Looks good to me; the only question I have is what is the change to
> mkinstalldirs for?

I have no clue what I did to change mkinstalldirs... it certainly should
not have been part of the patch in any case...

I still don't have a gnome--account, so what should happen now?  Should I
try to get first an account, and then patch, or do you want to commit the
patch yourself or should I first try to convert all other functions in
mp*.(c|h) to use this data-structure?

If you compile the patch, you will see that the compiler gives a looot of
warnings about pointers of wrong type... Many of these are easily
repairable.  They will just give some work, but they should be fine.  But
then there are some cases, where we will have to think a bit, because many
functions use the MP.t, MP.r etc. structures as a internal stack or as
some type of internal registers, and it is not always clear to me what to
do.  It seems that sometimes if the MPNumbers have length n, they want to
use for internal computations numbers of length n+2 to have more precision
or so...

Other than that, I still have a suggestion that we move all mp*-code into
a subdirectory.  This way, the core is really isolated from anything else
(and then I can maybe one day reuse it for the KDE-calculator ;-)).

Also, if there is somebody on the list that knows about yacc and lex, it
would be great if he/she could have a look at the unittests.  Run
"gcalctool -u".  The calculator does not handle "1e9" correctly, and I
wasn't able to figure out why that is...  About "Not 5", I believe to fix
these, one would just have to set the precision of the unittests higher...


Cheers

Klaus