Angus Leeming <[EMAIL PROTECTED]> writes:
| Lars Gullik Bj�nnes wrote:
>
>> Angus Leeming <[EMAIL PROTECTED]> writes:
>>
>> | Introduce it to the 1.4.x branch? Maybe. Slowly.
>>
>> boost::filesystem is not my Kartago.
>>
>> | But the real problem is that it is difficult to debug problems on
>> | a platform you do not use :-( In this case, I believe that the bug
>> | lies somewhere in the code in support/filetools.C's MakeAbsPath
>> | and indeed in support/tempname.C's tempName.
>>
>> yeah. and this seems to be the kindo of problems boost::filesystem
>> set out to solve.
>
| I'm not disagreeing with you. But it is not going to help us tackle
| the problem on the 1.3.x branch is it?
no. For 1.3.x any solution that does not change a lot of code is good
enough.
| What I think that we should do for the 1.4.x branch is use
| boost::filesystem for the internals of all the stuff in support and
| gradually change the interface to use boost::path rather than
| std::string.
>
| The problem is that we use this stuff all over the place. Do we want
| to make these changes before or after 1.4.x is released?
probably not, ... even if I have a tree largely changed to use
boost::filesystem...
but I have a feeling that boost::filesystem might be too strict.
--
Lgb
