Front page | perl.perl5.porters |
Postings from November 2000
Re: PerlIO - what next ?
Thread Previous
|
Thread Next
From:
Nick Ing-Simmons
Date:
November 9, 2000 02:06
Subject:
Re: PerlIO - what next ?
Message ID:
200011091006.KAA13894@mikado.tiuk.ti.com
Benjamin Stuhl <tiriath@yahoo.com> writes:
>
>P.S. PerlIO really has enough separate components it might
>be worth putting it into its own subdirectory.
Maybe. I really only has iperlsys.h and perlio.c right now.
But open.pm needs to be bundled too.
>It also
>might be worth making perlio.h not just be '#include
>"iperlsys.h"'.
Sure - once we have separately compiled layers then the struct
at least needs to be visible at least as an option.
This should probably happen very soon now.
The reason I did not do it straight away is that
perlio.h/iperlsys.h/perlsdio.h/perlsfio.h have got rather
entangled and I want to be sure that any bugs where in perlio.c
and not in the unravelling of those.
The reason it changed from perlio.h to iperlsys.h was for ActiveState's
"perl host" scheme. It makes some sense still as potentially at least
we need to be able to make "read/write/lseek" "virtual" on things
like Win32. We still need them for sysread() etc. but we don't
want them to call MS's (or Borland's) C runtime as that is where
the problems/differences lie.
>I'm starting to look at porting PerlIO to
>Win32
I had my ideas on that in mind while developing the stuff.
Please at least discuss your ideas with me.
Having someone else code it is most attractive, but I have a few
historical "gotchas" to pass on.
>and it would be much easier if each layer was its own
>source file (keeps me from having to create an #ifdef WIN32 morass).
If done "right" it should be "no worse" than the MMAP 'morass' in perlio.c
i.e. one #ifdef to enable/disable the define_layer and
a bigish block inside another #ifdef with the layer code.
But indeed it should be a separate file under win32/...
--
Nick Ing-Simmons <nik@tiuk.ti.com>
Via, but not speaking for: Texas Instruments Ltd.
Thread Previous
|
Thread Next