sunpro iterator test failures?
http://www.meta-comm.com/engineering/boost-regression/developer/iterator.htm l show mostly failures for the sunpro platform. These failures inhibit the serialization libraries from building - (and probably others as well). Are there any plans to accomodate this compiler? If not, should it continue to be tested? Robert Ramey
On Mon, 3 Jan 2005 09:24:47 -0800, Robert Ramey <ramey@rrsd.com> wrote:
http://www.meta-comm.com/engineering/boost-regression/developer/iterator.htm l show mostly failures for the sunpro platform. These failures inhibit the serialization libraries from building - (and probably others as well). Are there any plans to accomodate this compiler? If not, should it continue to be tested?
This compiler fails so many tests, turning it off might make sense. There are newer versions of SunPRO than the one I am using (I have CC: Sun WorkShop 6 update 2 C++ 5.3 Patch 111685-20 2004/03/19) which may be better, but I'm not in a position to upgrade. Anyone out there in Boost land have a newer version of SunPRO and care to run the regression tests? -- Caleb Epstein caleb dot epstein at gmail dot com
Caleb Epstein wrote:
On Mon, 3 Jan 2005 09:24:47 -0800, Robert Ramey <ramey@rrsd.com> wrote:
http://www.meta-comm.com/engineering/boost-regression/developer/iterator.htm l show mostly failures for the sunpro platform. These failures inhibit the serialization libraries from building - (and probably others as well). Are there any plans to accomodate this compiler? If not, should it continue to be tested?
This compiler fails so many tests, turning it off might make sense.
There are newer versions of SunPRO than the one I am using (I have CC: Sun WorkShop 6 update 2 C++ 5.3 Patch 111685-20 2004/03/19) which may be better, but I'm not in a position to upgrade.
Anyone out there in Boost land have a newer version of SunPRO and care to run the regression tests?
I do have access to the Workshop 8 / SunPRO 5.5 compiler at work and am planning to invest some time into trying to get more parts of boost to work with both the 5.3 and the 5.5 compilers. I'll see if I can find the time to run those tests tomorrow to see if they pass or not. Success so far has been rather limited, mainly due to the fact that "making boost work" isn't part of my real job description :-), however I've now got at least the smart pointer library working properly with both 5.3 and 5.5. More bits and pieces will hopefully follow over time; This depends somewhat if my employer would be kind enough to provide me with access to the compiler on my home SUN as I don't fancy spending too many nights in the office... HTH, Timo
On Mon, 03 Jan 2005 22:08:05 +0000, Timo Geusch <timo@unix-consult.com> wrote:
I do have access to the Workshop 8 / SunPRO 5.5 compiler at work and am planning to invest some time into trying to get more parts of boost to work with both the 5.3 and the 5.5 compilers. I'll see if I can find the time to run those tests tomorrow to see if they pass or not.
Thanks for taking the time
Success so far has been rather limited, mainly due to the fact that "making boost work" isn't part of my real job description :-), however
Nor is it mine. I'm just doing my best to help out by running the regression tests daily.
I've now got at least the smart pointer library working properly with both 5.3 and 5.5.
I'd love to have a copy of that patch if you wouldn't mind sending it along to the list.
More bits and pieces will hopefully follow over time;
I look forward to it.
Caleb Epstein writes:
On Mon, 03 Jan 2005 22:08:05 +0000, Timo Geusch <timo@unix-consult.com> wrote:
I do have access to the Workshop 8 / SunPRO 5.5 compiler at work and am planning to invest some time into trying to get more parts of boost to work with both the 5.3 and the 5.5 compilers. I'll see if I can find the time to run those tests tomorrow to see if they pass or not.
Thanks for taking the time
Success so far has been rather limited, mainly due to the fact that "making boost work" isn't part of my real job description :-), however
Nor is it mine. I'm just doing my best to help out by running the regression tests daily.
Yep, simply running the regression tests regularly is already a great service to the community. -- Aleksey Gurtovoy MetaCommunications Engineering
Caleb Epstein wrote:
On Mon, 03 Jan 2005 22:08:05 +0000, Timo Geusch <timo@unix-consult.com> wrote:
I do have access to the Workshop 8 / SunPRO 5.5 compiler at work and am planning to invest some time into trying to get more parts of boost to work with both the 5.3 and the 5.5 compilers. I'll see if I can find the time to run those tests tomorrow to see if they pass or not.
Thanks for taking the time
That's fine - we're all working towards the same goal...
Success so far has been rather limited, mainly due to the fact that "making boost work" isn't part of my real job description :-), however
Nor is it mine. I'm just doing my best to help out by running the regression tests daily.
I've now got at least the smart pointer library working properly with both 5.3 and 5.5.
I'd love to have a copy of that patch if you wouldn't mind sending it along to the list.
Will do - the main effort to get these to work was that I needed to find the correct combination of compiler flags to circumvent some of the braindeadedness - sorry, backward compatibility - of the SUN compilers. No changes to the actual code are required. However it does appear that most other libraries require patches to the code in order to compile.
Caleb Epstein wrote:
On Mon, 03 Jan 2005 22:08:05 +0000, Timo Geusch <timo@unix-consult.com> wrote:
I do have access to the Workshop 8 / SunPRO 5.5 compiler at work and am planning to invest some time into trying to get more parts of boost to work with both the 5.3 and the 5.5 compilers. I'll see if I can find the time to run those tests tomorrow to see if they pass or not.
Thanks for taking the time
They didn't pass - I've got the compiler output somewhere and will try to email them to the list in a separate mail, unless it is preferred if I stick it on some web space. It's fairly small (about 20k compressed). The error message are rather weird and need further investigation. I've managed to come to an arrangement with my manager that will allow me to use my work computer in my spare time to look into these problems and hopefully be able to track down the issues and submit fixes.
I've now got at least the smart pointer library working properly with both 5.3 and 5.5.
I'd love to have a copy of that patch if you wouldn't mind sending it along to the list.
It's actually not "patchworthy" at the moment, but the secret to success in this case is that you need to specify the following options: -features=tmplife This instructs the compiler to use standard C++ lifetime/scoping rules for temporaries. By default, it's backward compatible with old SUN compilers that implemented different rules and this breaks the boost smart_ptr tests. This is the only compiler flag you need to set on the SUNPRO 5.5 compiler. -instances=static The smart pointer templates and the SUN template cache database of the older compilers don't agree, which is why you're getting link failures when trying to link anything that uses them. This switch sticks all the static variables into the object file itself (it basically disables the template cache). Results are bigger object files, but together with the switch above this makes the smart pointer library work on the 5.3 compiler also. On Solaris 2.6 I also had to edit the jamfile to link it libposix4 for the real time functions, otherwise the tests themselves won't link. May be different on newer versions - I didn't have the time to verify that. Regards, Timo
On Thu, 06 Jan 2005 19:38:17 +0000, Timo Geusch <timo@unix-consult.com> wrote:
-features=tmplife This instructs the compiler to use standard C++ -instances=static The smart pointer templates and the SUN template
These features both appear to work (and to help) on SunPRO 5.3. I'm incorporating them into my regression test configuration. -- Caleb Epstein caleb dot epstein at gmail dot com
participants (4)
-
Aleksey Gurtovoy -
Caleb Epstein -
Robert Ramey -
Timo Geusch