Re: [bbv2] path.make for non-native paths? Re: [jamboost] Re: Documentation build broken?

From: Rene Rivera (grafik.list_at_[hidden])
Date: 2005-10-06 14:52:10

Vladimir Prus wrote:
> On Thursday 06 October 2005 07:11, Rene Rivera wrote:
> 
>>Rene Rivera wrote:
>>
>>>Eric Niebler wrote:
>>>
>>>>error: Unable to find file or target named
>>>>error: '/C:/boost/cvs/boost/boost/algorithm/string.hpp'
>>>>error: referred from project at
>>>>error: '../libs/algorithm/string/doc'
>>>>
>>>>That path with the leading '/' seems broken to me. Anybody else seeing
>>>>this?
>>
>>The documentation for path.make says that it "Converts the native path
>>into normalized form." But it is getting called in a variety of ways
>>with normalized paths. It wasn't a problem before because the path.join,
>>though the NORMALIZE_PATH and J="/" interaction, was swallowing this and
>>spitting out semi-reasonable paths out anyway. But now path.join is
>>somewhat less flexible. Volodya, or anybody who knows, is path.make
>>supposed to be used this way? Or, do I have to kludge
>>path.join/NORMALIZE_PATH to behave exactly as before?
> 
> path.make is supposed to be called with a path provided by a user, which can 
> be either in platform native format, or a relative path with forward slashes.

OK. I've kludge path.join to behave as before. And it fixes the problems 
with doc building.

> When it's called in other way?

The current issue is that path.glob returns normalized paths. So when 
glob is used within the declaration of a target those paths get the 
normalization treatment again.

> Maybe, I'll just need to go ahead and remove 
> all path.make/path.native calls completely, as Dave suggested several times.

Perhaps.

-- 
-- Grafik - Don't Assume Anything
-- Redshift Software, Inc. - http://redshift-software.com
-- rrivera/acm.org - grafik/redshift-software.com
-- 102708583/icq - grafikrobot/aim - Grafik/jabber.org