[Q] Re: [21.5] Implement FcConfig handling (95%)

[Aidan Kehoe]

 Ar an ceathrú lá is fiche de mí Aibréan, scríobh Stephen J. Turnbull: 

 > Aidan Kehoe writes:
 > 
 >  > I now have a Mac, so I’ll get to implementing the
 >  > Qfile_name-is-always-UTF-8 semantics there soon,
 > 
 > AFAIK this just involves adding another row to the table of platforms
 > with special needs in Mule initialization.  Don't *enforce* the
 > semantics of "always", because there are exceptions.

You mean people mounting Fat 32 drives with unspecified file name coding
systems? Or are there others? 

 >  > since the system normalises text as needed and we don’t have to
 >  > (though we should :-/ ).
 > 
 > What do you mean, we "should" normalize?  Normalization is an
 > optimization.  We should recognize all equivalent forms as equivalent,

To do that, we need to normalise. Recognising that "/tmp/aidan/äargh" and
"/tmp/aidan/äargh" are the same file can’t be done without it. 

 > but we can leave the actual normalization to those systems that do
 > normalization AFAICS.

Yeah, Apple have done a pretty sane implementation there, thankfully. 

 > We should be *able* to normalize for the purpose of communicating with
 > non-conforming systems, of course, but that is not trivial to
 > implement portably.

-- 
On the quay of the little Black Sea port, where the rescued pair came once
more into contact with civilization, Dobrinton was bitten by a dog which was
assumed to be mad, though it may only have been indiscriminating. (Saki)