Noticed a difference, is it between CentOS and Ubuntu or between GT.M and Yottadb?

[ed de moel]

I noticed an interesting difference...

I have two systems up and running.
The old one has CentOS, and is running GT.M V5.4-002B Linux x86_64
The new one has Ubuntu and is running GT.M V6.3-011 Linux x86_64 (YottaDB r1.34)

When I'm processing a multi-part payload from a POST request, on the "old" system, the line-ends come through as $Char(13,10), and on the new system they come through as $Char(13,0).
Does anyone know if that difference is caused by Ubuntu or by YottaDB?
Would this difference also impact image files that might be part of the multi-part POST?

Thanks,
Ed

[ed de moel]

Oh yes, if it matters:
The version of CentOS is: CentOS release 6.10 (Final)
The version of Ubuntu is: Ubuntu 22.04 LTS

[K.S. Bhaskar]

Ed, is this a SOCKET device? Also, do you have a small example of the M code? There have been changes in both GT.M and YottaDB between those releases to accept CR+LF as line terminators in various contexts, not just LF. However, I am not aware of anything that converts CR+LF into CR+null.

Regards
– Bhaskar

[ed de moel]

This is a web application, in other words YottaDB is invoked with the socket invoked by a webpage as its standard in/output.
The code just reads standard input with Read *byte as it is parsing the multi-part POST payload.

If you need more detail, please let me know.
Thanks.
Ed

[Sam Habiel]

Hey Ed,

It's completely possible that YottaDB/GT.M "fixed" something with delimiters somewhere along the years, but a $C(0) showing up is really strange. Is your client sending the data still the same?

I just noticed that you are reading standard input and output.

For reference, here's what I did in the M-Web-Server to deal with delimiters: https://github.com/shabiel/M-Web-Server/blob/master/src/_webreq.m. I delimit by $C(13,10) while reading the HTTP headers, and then switch to no delimiters while reading the data.

You said that you are using stdin/out. If so, you may find the code that is used here more useful, as it is invoked from xinetd: https://vivian.worldvista.org/dox/Routine_XWBTCPM_source.html

Let us know how you fare.

--Sam

[ed de moel]

Thanks, Sam. The "nodelimiter" setting did the trick!

Ed