Re:[boost] [Serialization] Remaining failures markup

Aleksey Gurtovoy wrote:
Robert,
Pending msvc-stlport issue aside,
I commented this accordingly.
the serialization library reports show quite a few failures with other toolsets, in particular:
cw-9.3 borland-5.6.4 cw-8.3 msvc vc7
I'm sure you've well aware of these; what I'd like to ask you is to comment on them in the "explicit-failures-markup.xml" in the similar way you commented on "*_warchive" tests, so that a) the library users are not scared away by all these yellow cells, and b) the release manager and everybody else besides the author have a simple way of telling whether the library is in the OK state or not.
I've been looking at this for a while procrastinated doing anything until I was sure what to do. So far this has been an effective strategy. Many times failures just disappeared as arcane tooI set issues were resolved. I still have some doubts about how to handle the remaining cases. Metrowerks compilers (cw*) seem to suffer from over-zealous optimization that eliminates static variables not referred to explicitly by name. Other compilers do this as well, but I've managed to workaround this in various ways. Cw* is the last hold out on this issue. I had hoped someone who is interested in this can test it might be motivated to look into it but it seems that there isn't enough interest. Unless I hear otherwise, I'll mark these tests as "expected". I would love to see this fixed - but for this Metrowerks would have an almost perfect showing. I do think its very high quality product that handles platforms that others don't. Its pains me that the serialization library test don't reflect this. A number of these (msvc, vc7, borland) have been determined to be manifestations of compiler deficiencies so I will mark them so. Some the borland test don't fail on my own test setup - so for me they are indeed "unexpected failures" - I think leaving this yellow is the correct characterization. That leaves a number of miscellaneous failures which really aren't understood or explained. Personally I would like to leave these as they are as to mey they really are "unexpected failures". I realize that some might be dissuaded from using the library because of this - but I think that's better than having someone use the library and not have it meet expectations. Robert Ramey

Robert Ramey wrote:
Aleksey Gurtovoy wrote:
Robert,
Pending msvc-stlport issue aside,
I commented this accordingly.
the serialization library reports show quite a few failures with other toolsets, in particular:
cw-9.3 borland-5.6.4 cw-8.3 msvc vc7
Metrowerks compilers (cw*) seem to suffer from over-zealous optimization that eliminates static variables not referred to explicitly by name. Other compilers do this as well, but I've managed to workaround this in various ways. Cw* is the last hold out on this issue. I had hoped someone who is interested in this can test it might be motivated to look into it but it seems that there isn't enough interest. Unless I hear otherwise, I'll mark these tests as "expected". I would love to see this fixed - but for this Metrowerks would have an almost perfect showing. I do think its very high quality product that handles platforms that others don't. Its pains me that the serialization library test don't reflect this.
I'd say mark them as expected for now. I can help in trying to resolve the link issues with CW but I don't have much time now. If you can give me an easy to reproduce example of the problem, hopefully outside of the serialization lib itself, it will make it much easier. I've had this problem myself and in one circumstance had to manually add references to exception classes in a dummy file+namespace to get their typeinfo linked into my DLL. -- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com - 102708583/icq

Robert Ramey wrote:
A number of these (msvc, vc7, borland) have been determined to be manifestations of compiler deficiencies so I will mark them so.
[...]
Robert (and others), please validate any xml files before you check them in. I've just commited a fix to make explicit-failures-markup.xml valid again. BTW: I suggest that everybody who alters any XML files does this using an XML editor! XML Spy is one of the best I know (for windows) and its Home Edition is freely available at http://www.altova.com. Stefan
participants (3)
-
Rene Rivera
-
Robert Ramey
-
Stefan Slapeta