>>>>> "Martin" == Martin v Loewis <martin@v.loewis.de> writes:
Thanks for the fast reply, Martin
[...]
>> Is the trial and error method the usual approach ? ;-)
Martin> No. Instead, one needs to clearly identify the problem. If
Martin> it is a compiler bug, it would fall into the class
Martin> "generates bad code". A useful bug report in this category
Martin> would identify a function, and present it together with
Martin> the assembler code.
Ok, I imagined sth. similar ...
>> Given the results above is it licit to blame the compiler(s)?
Martin> Not necessarily. It could be also that the program uses
Martin> undefined behaviour, such as accessing an uninitialized
Martin> variable. If that is a variable on stack, it might or
Martin> might not crash depending on the previous value of the
Martin> stack.
So, concluding, the following isn't correct?
,----
| #
| # If SWISH++ doesn't work correctly with optimization on, but it
| # works just fine with it off, then there is a bug in your
| # compiler's optimizer.
`----
Regards,
MH
--
(Dr.) Michael Hummel
mailto: mh@seitung.net || molino@gmx.net
--
fprint = F24D EAC6 E3D7 372C 9122 D510 EB24 01CA 0B56 B518
id: 1024D/0B56B518 key: http://www.seitung.net/keyAttachment:
pgpe4OLjD6gO6.pgp
Description: PGP signature