[Typeof] Formatting regression results
A suggestion to improve the display of the typeof results page: If you look at the boost iterator library, you will see the the 'corner cases' appear as a separate section. Apparently this can be achieved by tweaking the explicit-failures.xml markup file:
This is done with the category attribute of the test entity in the XML document describing the expected failures. For example:
<test name="indirect_iter_member_types" category="Corner-case tests"/>
I believe this page might be more helpful if 'emulated' and 'native' mode were split into two categories. I am happy to edit the XML file to make this work (I have commit rights) but I really need the library authors blessing before trying this! Arkadiy, Peder : what do you think? Any other library authors interested? e.g. tr1: std vs boost tests -- AlisdairM
2007/5/2, AlisdairM <alisdair.meredith@uk.renaultf1.com>:
A suggestion to improve the display of the typeof results page:
If you look at the boost iterator library, you will see the the 'corner cases' appear as a separate section.
Apparently this can be achieved by tweaking the explicit-failures.xml markup file:
This is done with the category attribute of the test entity in the XML document describing the expected failures. For example:
<test name="indirect_iter_member_types" category="Corner-case tests"/>
I believe this page might be more helpful if 'emulated' and 'native' mode were split into two categories.
I am happy to edit the XML file to make this work (I have commit rights) but I really need the library authors blessing before trying this!
Arkadiy, Peder : what do you think?
I think it is a great idea. You have my go ahead. Peder
Any other library authors interested? e.g. tr1: std vs boost tests
-- AlisdairM
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
"AlisdairM" <alisdair.meredith@uk.renaultf1.com> wrote
A suggestion to improve the display of the typeof results page:
If you look at the boost iterator library, you will see the the 'corner cases' appear as a separate section.
Apparently this can be achieved by tweaking the explicit-failures.xml markup file:
This is done with the category attribute of the test entity in the XML document describing the expected failures. For example:
<test name="indirect_iter_member_types" category="Corner-case tests"/>
I believe this page might be more helpful if 'emulated' and 'native' mode were split into two categories.
I am happy to edit the XML file to make this work (I have commit rights) but I really need the library authors blessing before trying this!
This is a good idea, but I assume you are talking about 1.35 :) Then you can go ahead. Also, I believe the typeof testing needs to be generally reviewed. Right now all the "native" tests repeat emulation tests just with different flag. I don't think this has a lot of sense, since the emulation tests were designed to test registration. Regards, Arkadiy
Arkadiy Vertleyb wrote:
This is a good idea, but I assume you are talking about 1.35 :) Then you can go ahead.
Absolutely! The last test page generation for HEAD has this update working now. My first attempt was ignored by the test page generator)
Also, I believe the typeof testing needs to be generally reviewed. Right now all the "native" tests repeat emulation tests just with different flag. I don't think this has a lot of sense, since the emulation tests were designed to test registration.
I'll worry about that after I get Borland emulation working better <g> Note that I have some uncommitted patches that visible in the current results page. -- AlisdairM
"AlisdairM" <alisdair.meredith@uk.renaultf1.com> wrote
The last test page generation for HEAD has this update working now. My first attempt was ignored by the test page generator)
Cool!
Also, I believe the typeof testing needs to be generally reviewed. Right now all the "native" tests repeat emulation tests just with different flag. I don't think this has a lot of sense, since the emulation tests were designed to test registration.
I'll worry about that after I get Borland emulation working better <g>
The current state of Borland emulation is better than I thought -- have you already committed anything? Since type_emulation and template_type_emulation are working, I think the library is already usable, since these two, I think, cover 90% of all usage. If you get template_int_emulation working, it will be 99%. The rest covers rather exotic cases.
Note that I have some uncommitted patches that visible in the current results page.
Do you mean "not visible"? Regards, Arkadiy
Arkadiy Vertleyb wrote:
The current state of Borland emulation is better than I thought -- have you already committed anything? Since type_emulation and template_type_emulation are working, I think the library is already usable, since these two, I think, cover 90% of all usage. If you get template_int_emulation working, it will be 99%. The rest covers rather exotic cases.
Great - I never know what the important cases are on librarieS I can't use - my goal is simply to get all the green lights <g>
Note that I have some uncommitted patches that visible in the current results page.
Do you mean "not visible"?
Doh! No, I meant the test results that are visible refer to the un-published patch. Note that the tests you see also apply to a non-public version of the compiler, although the current version passes most the same tests, including the two type cases you mention. I'll dig it out shortly and post it for comment - it is a pretty brute-force approach! -- AlisdairM
"Arkadiy Vertleyb" <vertleyb@hotmail.com> wrote
"AlisdairM" <alisdair.meredith@uk.renaultf1.com> wrote
Note that I have some uncommitted patches that visible in the current results page.
Do you mean "not visible"?
Oh, I see now... It's running on you machine, so you don't have to commit for the results to become visible, right? Regards, Arkadiy
participants (3)
-
AlisdairM -
Arkadiy Vertleyb -
Peder Holt