[Aldor-l] Compilation problem

[Ralf Hemmecke]

Hi Laurentiu,

> Try with AXL_EDIT_1_1_13_18 set to 0. Rebuild everything from
> scratch. It fixed the issue on my machine. This fix was enabled to
> fix a regression test that was failing. Unfortunately, it breaks some
> tests in other libraries too, so I guess we should leave it as
> disabled and find a proper fix.

Thank you for this hint. That makes the issue I had go away.

But it looks *very*, *very* strange. It looks like a bugfix from the NAG
days. So if I now have to switch it off then is it right that *you* also
switched it off the binaries

http://www.aldor.org/downl.html
Linux version 1.1.0 (glibc2.3) (this is a release candidate)

??? Maybe it has always been switched off for any distribution of the
Aldor 1.0.x series?

Are there some other changes to the sources I have to apply to be able
to actually produce the same executable as would be distributed by
aldor.org?

It would be very helpful, if all such issues are made public. There is
no shame in showing all of the current bugs. Aldor is buggy and we all
know it. But if all the old bugreports are kept secret, then other
people cannot easily work on them. Could you enter the old bugreports
into the bugtracker?

Ralf

PS: If I look at editlevel.h it says...

---quote
The problem with Basicmath DUP can be seen when * is applied to acc and
result in CantorZassenhausFactorizationPackage (located in 
$BMROOT/src/polypkg/factorczpkg.as).  When acc is ? and result is 1,
(both acc and result being of type U), the multiply invoked is (r: R) * 
(x : %) : % rather than (x : %) * (y : %) : %.  This results in a large 
integer times ? being returned, rather than 1*?.
---endquote

Does that mean that with AXL_EDIT_1_1_13_18 set to 0, I will get the 
problem that

   *: (R, %) -> %
and
   *: (%, %) -> %

will not work properly?