[Info-vax] RMS internals?

[JF Mezei]

Bob Koehler wrote:

>    So you're skipping RMS and trying to rewrite it yourself?  Sounds
>    self defeating.  Anything you can get that way that you can't get
>    by increasing the buffer size?

The way I understand it, his data consists of records greater than 32k.

As I understand it, the "length" is a signed 2 byte value (is that
correct) which makes the record size limit 32k. So his records cannot be
represented as variable length records.  The FTP server thus splits them
into 2 records when transfered as text (which generates variable length
records).

Can the FTP server be told to create text files as stream/stmlf instead
of variable length ? That might be a solution.

The other solution is to have the files transfers binary, preserving
whatever line termination they had on windows, then use SET
FILE/ATTR=RFM=STMLF (or whatever "stream" variation matches that of
windows). At that point, RMS should be able to read the records,
assuming buffquota is big enough.

But if the "size" argument of a descriptor is also just 2 bytes and they
are signed, then he will have the same problem. (It's been a while since
I looked at the descriptor stuff, is the size 2 or 4 bytes ?).