On Tue, Feb 08, 2005 at 03:08:29PM -0700, Jonathan Turkanis wrote:
Hi Jonathan,
On Tue, Feb 08, 2005 at 01:39:26PM -0700, Jonathan Turkanis wrote:
VC7.0: passes no tests, but same was true prior to the changes, on my installation (although the regression reports show three tests passing)
Let me just comment this issue. I have created a workable core for the vc7.0 It was used by the string_algo internaly until I have switched to
Pavol Droba wrote: the
Boost.Range.
I was discussing the issue with Thorsten, and as the result I'm looking for a way how to integrate my workaround code into the current version of Boost.Range.
However, I have no access to vc7.0 compiler so I will take a while until I will finish it. If you have access to it and some time to spare, it could help a lot.
Well, I don't have too much time to spare, but perhaps it won't take long. Should I start with algorithm/string/collection_traits.hpp and algorithm/string/iterator_range.hpp, rev 1.6, or do you have a more recent version?
According to my observation, it seems, that current workaround code, located in range/detail/common.h is not working properly. This code is supposed to qualify the kind of range being used. The qualification seems to be failing. I have developed different scheme, and it is located in the collection_traits* files. Unfortunately, this files are no longer compatible. collection_traits* was designed for the older version of range library. So, the code has to be ported to make it usable. To make thing easier, it should be enough if the array recognition code is ported, since the rest of the stuff is generaly the same. However, it is still possible, that the current code is generaly okey, just it needs some final touch. Maybe this is the best way you can go. If you'll be able to localize the problematic spot, we can either fix it, or rework it. Regards, Pavol