On Mon, 9 Jun 2008, Roland Mainz wrote:
> Kuldip Oberoi wrote:
>> Roland Mainz wrote:
>>> Sean Sprague wrote:
>>>>> Does anyone know when (approximately) the Sun Studio version after
>>>>> "Studio 12" will be released ?
>>>>>
>>>> I guess that the obvious question is "what extraq functionality do you
>>>> need? Cross-compilation?"...
>>>
>>> In general: More bugfixes, working "dbx -check access" (SS11's "dbx
>>> -check access") is currently defunct in Solaris 11/B84), better XIPO
>>> support, builtin checks for thread race conditions and some other
>>> stuff...
>>>
>>>> I suspect that you will hear from Kuldip soon.
>>>
>>> Erm... when ?
>>
>> There's not a set date, yet, for the next FCS release of Sun Studio, but
>> we are planning on releasing the next Sun Studio Express build within
>> the next 6-8 weeks (roughly).
>>
>> Access checking did improve with SS12 + latest patches (SS12 has support
>> for x86), but if there are issues/RFEs, let us know. What improvements
>> are you specifically looking for in XIPO?
>
> Erm... it should simply work (and not crash).
file bugs, fule bugs, file bugs.
>
>> Good time to get RFEs/etc.
>> in!
>
> One RFE would be a message system when/if the compiler does any
> optimisations (and which one). I strongly miss the (Amiga) SAS/C feature
> which was announcing informations like (you already have something like
> this for loop paralisation (e.g. -xloopinfo)):
> - Variable removed
> - Variable datatype changed (e.g. local |int| reduced to |bool| when
> only { 0, 1 } were used as value (and no pointer to this variable was
> used))
> - Code block removed, moved etc.
> - Function inlined
> - Function declared inline but not inlined
> - Loop unrolled
> <...and so on...>
>
> Such a system could greatly help finding bugs in the optimizer or at
> least guess what the compiler was doing...
man er_src, but in general I agree, it is not the best way to do it.
> 2nd RFE: Enable "-xcsi" by default (and provide an option "-xnocsi" to
> turn it off). The compiler should be CSI-conformant by _default_ (note:
> This does _not_ mean localisation (= l10n), I am talking about multibyte
> characters in the source code) and not optionally. gcc supports this
> since _years_ and it's one of the "sour points" for non-english speaking
> users like the japanese/chinese developer communities (it's not even
> well known that Sun Studio acutally support this... ;-( ).
>
>> Also, there is runtime analysis of race conditions via Thread
>> Analyzer in Sun Studio 12. Are you looking for enhancements?
>
> Well... a scriptable interface would be nice (e.g. something which
> doesn't require a X11 connection).
man er_print
man collect
>
> ----
>
> Bye,
> Roland
>
> --
> __ . . __
> (o.\ \/ /.o) roland.mainz at nrubsig.org
> \__\/\/__/ MPEG specialist, C&&JAVA&&Sun&&Unix programmer
> /O /==\ O\ TEL +49 641 7950090
> (;O/ \/ \O;)
> _______________________________________________
> tools-compilers mailing list
> tools-compilers at opensolaris.org
>
