[boost] Re: Problem with filesystem attributes

From: Jan Langer (jan_at_[hidden])
Date: 2003-06-30 14:16:16

Fanta, Richard wrote:
 > I've been attempting to try out the attributes code in the boost-sandbox,
 > but am running into some snags using VC++ version 7 (that comes with
 > VS .NYET).

i am really glad that your are looking at this code.
(un)fortunately i dont have vc++ available and did not implement the
attributes for vc++.
the stuff in the sandbox implements _some_ attributes on a posix platform.

after writing this interface (based on previous work of dietmar kuehl),
testing it on a few simple examples and some short discussions here, i
realized that there is a lot more work needed. like finding a
appropriate set of attributes, implementable on most platforms and at
least an implementation for vc++.

 > One simple goal here would be to discern whether a file is writeable.
 > Can this be done with the filesystem::attributes library??

not at the moment. but it is not so easy to say when a file is writeable
and so this attribute did not belong to my small testiung set of
attributes. but one of my main goals when writing it was the possiblity
to add user defined attributes easily.
(maybe one should regard the attributes lib at the moment just as a
framework for your own attributes.)

 > The code for attr_test.cpp from the tests directory is included below;
 > it's been slightly modified to run stand-alone instead of through the
 > boost test harness.
 >
 > Problem #1: No string() method on the cache class.
 >
 > Problem #2: The "<<" operator won't work.
 >
 > For the latter, I've seen it suggested that typedefs might help, but
 > don't understand what the underlying issue is. Is this a lingering
 > Microsoft (i.e. VC++) specific issue? I'd greatly appreciate any
 > clarification that people can offer.

with gcc there are no such problem. it looks like something, which can
be solved by some additional parenthesis. but even if the syntak is
correct it will not work on a non-posix system because nothing is
implemented.

actually the get function takes a cache (pm) and returns a path on which
string() is called.

jan

-- 
jan langer ... jan_at_[hidden]
"pi ist genau drei"