From: Jürgen Hunold (hunold_at_[hidden])
Date: 2006-04-05 03:35:41
Am Middeweken 05 April 2006, 09:17 schrieb Vladimir Prus:
> until today, intel-win.jam passed a number of frontend options to compiler,
> When trying to run regression tests with Intel-Win 9.0, Stefan Slapeta run
> into internal compiler errors. Removing the above options fixes that.
> They were added by Jurgen's suggestion, who said they disable "microsoft
> bug emulation", IIRC. On the other hand, Stefan said this:
Yes. Thomas Witt figured those out :-) The last intel compiler used by us were
6.0 and 7.0 AFAICT.
> -Qoption,cpp,--const_string_literals ... that should be the case anyway
> -Qoption,cpp,--new_for_init ... why? there's a documented option for
> this -Qoption,cpp,--no_implicit_typename ... same as (1)
> -Qoption,cpp,--no_friend_injection ... why?????
> -Qoption,cpp,--no_microsoft_bugs ... that's very dangerous as intel have
> their own mode for microsoft compatibility. the std library would
> compile without that
Well, intel 6.0 was bug-compatible with msvc-6.0 by _default_, so those
switches were needed to enable the full power of the EDG-Frontend ;-))
As msvc 7.1 and 8.0 are more standard compliant, it might be the reverse right
now ;-)) and EDG is running with all features enabled by default...
> I'm not an expert with Intel compiler, so have no idea what options are
> really necessary at the moment. Can somebody comment?
I'd suggest to keep this options for intel < 9.0 (or 8.0) until someone with
more knowledge comes with a better solution.
Just what I remember :-))
-- * Dipl.-Math. Jürgen Hunold ! Ingenieurgesellschaft für * voice: ++49 511 262926 57 ! Verkehrs- und Eisenbahnwesen mbH * fax : ++49 511 262926 99 ! Lister Straße 15 * hunold_at_[hidden] ! www.ive-mbh.de
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk