Bjørn Roald wrote:

Is this a download available to the public? Yes, the SunStudio 11 compiler is freely downloadable from Sun's website.

Sun finally ditched the whole licensing the compiler thing (seems trendy, since now you can download the MSVC compiler and Borland compiler (without the IDE) for free too). I guess they were seeing too many people using gcc on solaris as an alternative. Either way, this is a Good Thing (tm). Though unfortunately their STL support isn't that great in Sun Studio 11, however the bundle STLPort with it, and its easy enough to use STLPort instead of their native STL (its a simple compiler flag), which I believe boost's old SunStudio support already has taken care of.

Hmmm - are there work-arounds for known compiler bugs activated in boost config when you set sunpro? If there are, they may be worth looking into. I am saying this because I am hopeful Sun are fixing their compiler. This is based on statements that they intend to support Boost with their compiler - and they do not seem to participate in efforts to add workarounds for their bugs in Boost.

I played with this myself at my last job. I got a lot of it working with some simple changes to things such as is_enum.hpp as mentioned in the parent. Most of the changes I had to do were simply using the compiler workarounds in use for other compilers, occasionally it required me to do something special, but not very often and not too complex. Granted, my goal was to get program_options working on SunStudio 11, so once I had achieved that, I stopped. So I didn't scour through all of boost, but I'd say that SunStudio 11 (with STLPort) is actually now ready to be a supported compiler by Boost if someone will actually go in and add the appropriate #ifdef's to activate the correct compiler behavior. Right now my SPARC system is not really in a state I can play with this (its also only a SPARC Ultra 5, 300mhz with 256mb ram, so probably not really a good system to compile boost on, either ;) Otherwise I might play a little more around with it now I can get the compiler for free. I'd love to use the native Sun compiler on Solaris systems. -- PreZ :) Death is life's way of telling you you've been fired. -- R. Geis

Harri Hakula wrote:

On 1/15/06, Preston A. Elder <prez@neuromancy.net> wrote:

...

Otherwise I might play a little more around with it now I can get the compiler for free. I'd love to use the native Sun compiler on Solaris systems.

The Sun compiler team is actively working on getting boost up and running. This is the most recent thread on their forums http://forum.sun.com/thread.jspa?threadID=28265&tstart=0

Quote from the most recent message on the thread: "Yesterday finaly managed to get the newest Boost (1.33.1) compiled with the SUN Studio 11 compiler! This was after installing the newest patches for the C++ compiler (16-dec-2005)."

Unfortunately there is no contact information for the poster "Henry."

I'd like to hear from boosters with access to Studio 11 with all the latest patches.

Silly me, I didn't realise that there were already patches for Studio 11. I've just checked and there are indeed patches available. I'm currently running on the vanilla compiler but I'll see if I can install those patches fairly soon.

On Sunday 15 January 2006 17:38, Timo Geusch wrote:

Bjørn_Roald wrote:

...

On Sunday 15 January 2006 10:02, Timo Geusch wrote:

...

So far, the overall results look a lot better compared to Studio 8 - after a build of the 'state' target with the basic sunpro tool set and absolutely no changes to the existing 1.33.1 configuration, bjam reports the following: ...failed updating 360 targets... ...skipped 192 targets... ...updated 877 targets...

the numbers seem to be low to me, I get more than 9000 targets for all of boost. Do you try with all of boost or only parts?

I'm trying the basic bjam "-sTOOLS=sunpro" stage at this point. I don't have Python on this box yet so I won't be able to build the Python stuff initially but I'll sort this out when I'll get around to it.

Ok, I also miss-read your post saying <quote> 'state' target </quote> to mean that you where running bjam in the "status" directory. :-) --- that is where I get 9000 + targets. I think building in that directory run most, if not all the test as well as building boost. Bjørn

Beman Dawes wrote:

"Timo Geusch" <timo@unix-consult.com> wrote in message news:1137315728.732.14.camel@sparcy...

...

regular regression test and provide the results but I'm probably getting a tad ahead of myself...

I'd really like to encourage you to start posting regression results for the SunStudio 11 compiler.

I will be. It's a matter of time for me - like just about everybody else on this list I've got a day job as well to pay for toys, err, the mortgage... My current plan is to get the compiler config sorted out to the point where it builds as much of Boost as I can get done within a week or so. Once that's done then I'll get cracking on the regression testing - as it's my box there's nothing to stop me from setting up a nightly cron to run the tests against CVS-HEAD, but I'm not there yet. That said, one of the reasons I got this box here was to do exactly that.

Steve Clamage at Sun has said he would like to see Sun doing better with Boost code, and they are doing some internal testing to ensure this. But there is nothing as good as regular posted regression tests for boosters to work with. Understood.

There is a mailing list (see http://lists.boost.org/mailman/listinfo.cgi/boost-testing) just to help folks wishing to contribute regular test runs. The folks that follow that list are very helpful in getting set up to do regular tests.

Thanks, I'll subscribe there once I take the first hurdle. Regards, Timo