Hi Gennadiy, Am Montag, 5. Oktober 2015, 15:08:38 schrieb Gennadiy Rozental:

Jürgen Hunold writes:

...

-- is required before extra test module arguments

See my post here:

http://article.gmane.org/gmane.comp.lib.boost.devel/263485

Thanks, I missed that one. But iz should have been directed to the Boost.Build mailing list as well. And it is not only Boost.Test which require changes, there are several testers broken, e.g. http://www.boost.org/development/tests/develop/developer/output/teeks99-08m-... Adjusting each test Jamfile does just not scale, we need a more global solution which requires changes to testing.jam itself. And last but least: Why do you require argument separation at all? I'm missing a rationale for this decision. (Or just overlooked it again...) Yours, Jürgen -- * Dipl.-Math. Jürgen Hunold ! * voice: ++49 4257 300 ! Fährstraße 1 * fax : ++49 4257 300 ! 31609 Balge/Sebbenhausen * jhunold@gmx.eu ! Germany

Hi Gennadiy, Am Montag, 5. Oktober 2015, 18:47:20 schrieb Gennadiy Rozental:

I tried to explain rationale here:

http://article.gmane.org/gmane.comp.lib.boost.devel/263463

Gah, I overlooked that one, too. The rationale makes sense.

In a summary: we want to properly report incorrect command line arguments. This kind of scheme is used in many places and Boost.Test just follow the suite.

I'm with out at this. Maybe some more SHOUTING in the subject would have been better. And maybe some preprocessor #define to select old/new commandline parser during compilation. This would have allowed the testers to use the old behaviour while figuring out how to adapt testing.jam. The same goes for c++11/03 switches, of course. But I think that going forward is the best way out of this... Yours, Jürgen -- * Dipl.-Math. Jürgen Hunold ! * voice: ++49 4257 300 ! Fährstraße 1 * fax : ++49 4257 300 ! 31609 Balge/Sebbenhausen * jhunold@gmx.eu ! Germany

Hi Raffi, Am Donnerstag, 8. Oktober 2015, 09:23:49 schrieb Raffi Enficiaud:

Le 07/10/15 18:15, Jürgen Hunold a écrit :

...

I'm with out at this. Maybe some more SHOUTING in the subject would have been better. And maybe some preprocessor #define to select old/new commandline parser during compilation. This would have allowed the testers to use the old behaviour while figuring out how to adapt testing.jam. The same goes for c++11/03 switches, of course. But I think that going forward is the best way out of this...

I reverted the changes, we will implement it slightly differently.

Okay. Just ping me if you need some Boost.Build support. I can't offer much, but at least I've got something working for the Boost tree. See https://github.com/boostorg/build/tree/test_args for the feature branch of the "--" spearator feature. It needs some finetuning and needs to be configurable, but works at least with Boost.Test in the Boost tree. Yours, Jürgen -- * Dipl.-Math. Jürgen Hunold ! * voice: ++49 4257 300 ! Fährstraße 1 * fax : ++49 4257 300 ! 31609 Balge/Sebbenhausen * jhunold@gmx.eu ! Germany