On Mon, 25 Jan 2010 10:40:35 -0500, P S wrote:
>On Mon, Jan 25, 2010 at 10:33 AM, Miklos Szigetvari
><[email protected]> wrote:
>> I write now via "Say".
>> As I set the HTML "content-encoding" to binary , the transer is o.k. so the
Binary? As in application/octet-stream, which is what I'm accustomed to?
>> PDF binary transsfered, but as I'm using the "Say" to write the HTML I got
>> extra newlines .
>> I would like to get rid of this newlines
>
>Ah. A SAY generates a newline, so you need to build a huge, ugly
>record and SAY that. Not sure if that gets into limitations on SAY
>length; some experimentation with non-binary long lines is probably
>indicated.
>
I think the limit is about 1000. Used to be terminal line width,
defaulting to 80 if no terminal was attached. They fixed it,
sort of. But:
o If there's any limit at all, for consistency it should be LINE_MAX
which is about 2000, IIRC.
o There should be no limit whatever under Unix System Services --
the argument of SAY should simply be streamed to stdout with
a newline appended. It appears Rexx misuses a layer intended
for unit record devices to address stdout. Just plain stupid,
not rocket science.
Instead of SAY, use:
RC = SYSCALLS( 'ON' )
address SYSCALL write ...
My experience has been that I can switch descriptorss behind Rexx's
back with no buffering latency. But experiment; RTFM for
specifications; submit RCF because there aren't any.
-- gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
