Re: [Sbcl-devel] non-ascii alien c-string and pathnames

[Christophe R. <cs...@ca...>]

Yaroslav Kavenchuk <kav...@je...> writes:

> Many thanks for response.
>
> One question:
>
>> [ As a stylistic issue, the patches currently have too many sb-unicode
>>   reader-conditionals. For example, instead of having
>> #!-sb-unicode 'simple-base-string  #!+sb-unicode 'simple-string
>> I'd prefer either just using:
>>     'simple-string
>>   or:
>>     (deftype foo-string ()
>>       #!-sb-unicode 'simple-base-string       #!+sb-unicode
>> 'simple-string)
>>     [...]
>>     'foo-string
>
>
> What it is more preferable?

If simple-string is the right thing for a build with #!+sb-unicode, it
is likely also to be the right thing for a build with #!-sb-unicode.
The reason is that simple-string includes all array specializations of
subtypes of character, including the somewhat pathological
(simple-array nil (*)).  If including that is right with sb-unicode,
it is also right to include it without.

Cheers,

Christophe