Gil wrote:

>OK.  I'll try.  Simplicity of specification.  Simplicity of implementation.

>Filenames are strings.  Different strings should refer to different files.

 

Categorical imperative there. Seems…circular.

 

>Consistency.  With Binder it's easy enough to create a load module:

>    CASE(M)

>        ....

>    NAME  FooBar(R)

 

>Should //STEP EXEC PGM=FOOBAR  invoke that program?  Why not"

>How about //STEP EXEC PGM='FooBar'?  Why not?  How about

>TSO:  EXEC *(FooBar)?

 

Binder? I’m talking Unix going back 45 years, not USS. But yes, I’d expect
them to invoke the same program.

 

>Would you submit or vote for an RFE that LOAD/LINK/ATTACH, BLDL, ...

>be made case-insensitive?

 

Yes.

 

>Why not?  I suspect you supplied the answer:

 

>>So it fits the definition of "tradition": The same stupid old way we've

>>always done it!

>>  

>"Stupid" indeed.  And z/OS is worse than most for inconsistency.  Some

>interfaces are case-sensitive; others enforce case-insensitivity.

 

Absolutely z/OS is horrible in this regard. So?

 

>And ethnic diversity.  Should files named in Cyrillic, Greek, ... be
treated

>in a case-insensitive fashion?  Imagine the implementation complexity

>and documentation complexity.  Should it be locale-sensitive?  Should

>Cyrillic filenames be case-insensitive in the Russia locale and Latin

>filenames be case sensitive?  And vice-versa in a Latin locale?

 

Well, going back to the early days of Unix, I don’t think any of this
mattered, so it’s not a defense for the design. But why would
case-sensitivity need to change across locales?

 

>Suppose another language is newly added to the Unicode CECP.  Should

>characters previously considered distinct suddenly be considered equivalent

>because they are upper-lower case pairs?

 

>(Don't be Anglocentric in your answer.)


Are you suggesting that there are codepoints that appear in multiple pages
but map differently—so “a” and “O” might be the upper/lowercase versions of
the same character in Blezerbian? I respectfully disbelieve that.

 

>Others have argued here that the filesystem should ignore diacritical
marks.

>But a Hispanophone sees "año" and "ano" as two very different nouns and

>would probably not approve of using them interchangably as a filename.

 

Those are different characters. Unicode folding is an entirely different
issue, I think.

 

>Peter Relson, among others, has written here of "invalid" names, implying

>GIGO.  I disagree with quiet GIGO -- a programmer should be provided at

>least a warning message on use of an "invalid" construct.

 

Sorry, don’t grok this point at all!

 

Cheers,

…phsiii


----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to