Re: RFC 14 (v2) Modify open() to support FileObjects and

2000-08-07 Thread Nick Ing-Simmons
lass accordingly - i.e. if you open a disk file you get something IO::File-ish, if you open a directory you get something IO::Dir-ish, a named pipe you get something where seek() either dies or does sfio-like fake seek of shadow buffer and a "tty" could have extra methods to give access to st

Re: RFC 14 (v2) Modify open() to support FileObjects and

2000-08-09 Thread Nick Ing-Simmons
tory (note the trailing '/') > >Portability, not all platforms use / Allowing URI/URL syntax can perhaps dodge the "portability" issue URLs use / so if user thinks URL they think /. -- Nick Ing-Simmons

Re: RFC 14 (v2) Modify open() to support FileObjects and

2000-08-16 Thread Nick Ing-Simmons
me - what are (hopefully) indistinguishable from native in the UNIX case at least. -- Nick Ing-Simmons

Re: RFC 14 (v2) Modify open() to support FileObjects and

2000-08-16 Thread Nick Ing-Simmons
m, $args; should "just work" if the module is in the right place, and give a meaningful error if it isn't. > > open MyHttp, 'http://www.yahoo.com/', $custom, $args; > >Allows you to do this simply, in a syntax that is already established. The URI syntax is also well established. -- Nick Ing-Simmons

Re: RFC 30 (v2) STDIN, STDOUT, and STDERR should be renamed

2000-08-16 Thread Nick Ing-Simmons
ified it does NOT change STDOUT. It will be impossible to translate many perl5 programs to perl6 if that difference is not retained somehow. -- Nick Ing-Simmons

Re: RFC 30 (v2) STDIN, STDOUT, and STDERR should be renamed

2000-08-16 Thread Nick Ing-Simmons
probably mention here that the single-arg form of select() is the >one you're suggesting for deprecation, and not the four-arg form. The 4 arg form will be deprecated somewhere else. Splitting the function is a good idea... -- Nick Ing-Simmons

Re: RFC 30 (v2) STDIN, STDOUT, and STDERR should be renamed

2000-08-16 Thread Nick Ing-Simmons
shift; print Dumper($arg); print STDOUT "Dumped ",ref($arg),"\n"; } select(STDERR); show_things({a => 1}); open(my $foo,">dump"); select($foo); show_things({b => 2}); close($foo); select(STDOUT); show_things({c => 3}); __END__ This technique is extremely handy! Now if only one could divert warn/die messages to STDERR by a similar trick -- Nick Ing-Simmons

Re: RFC 69 (v2) Standardize input record separator (for

2000-08-16 Thread Nick Ing-Simmons
s is almost certainly NOT a global 'use' option but a "mode" of a particular handle. We want $/ to go away and become a property of the handle. Perhaps $/ or the 'use' can set the _default_ which is passed to the "open" when nothing more specific is given. > >-Hao -- Nick Ing-Simmons

Re: Internal Filename Representations (was Re: Summary of I/O related RFCs)

2000-08-16 Thread Nick Ing-Simmons
n other words: things like >the protocol, the port number, the username, the password, could >be part of a "file spec". Quite. -- Nick Ing-Simmons

Re: Internal Filename Representations (was Re: Summary of I/O related RFCs)

2000-08-16 Thread Nick Ing-Simmons
, IOSUBSYS ]. >The file name is RMM, the extension is PDR. > >>$fo = open "/etc/inet/inetd.conf"; >>$fo->pathdrive = ""; > >I think this should be the mount point, e.g., "/". > >> Splitting apart or putting together either one of thes

Re: PerlIO - what next? - (should getc get a character?)

2000-11-04 Thread Nick Ing-Simmons
Uri Guttman <[EMAIL PROTECTED]> writes: >>>>>> "NI" == Nick Ing-Simmons <[EMAIL PROTECTED]> writes: > > > > NI> I have guts of a stack-of-layers PerlIO scheme coded now > NI> (//depot/perlio/... for those with perforce access - merg

PerlIO - what next? - (should getc get a character?)

2000-11-04 Thread Nick Ing-Simmons
than read (cf sfio). "win32" layer that does IO straight to Handle level rather than via MS's idea of how UNIX read/write work. -- Nick Ing-Simmons