twf-fix-compress-again2 is no good about mdsplus HOT 6 CLOSED

Can you come up with a simple test that demonstrates this? I can't figure out how the change could be causing a segfault. I've tested this on fc22 with both floating point arrays and word arrays and see no problem.

The jas-fix-compress-error-for-FS-minus-zero fix essentially disabled #198 so it restored the correct compression with earlier gcc versions (rhel6) but also restored the segfaulting behavior with more recent gcc versions which that fix was attempting to correct. This latest fix caused the code to behave consistently on the various versions of the compiler whether they were compiled with -O2 or -O0 and should have restored the behavior of compression which has been working successfully for decades.

from mdsplus.

I was able to reproduce the segfault using an ACQ196 in my office with a different server running 7.0.206 -- initftp, provide external trigger, store. Reverting back to 7.0.205 fixes it. I switched back and forth between the two versions a few times just to be sure. I can provide the test model later in the evening if it helps.

from mdsplus.

Can you set the nodes under the device to no compress store it and send us the tree ? That way we can read the data and try to write it into a compress on put node
This should allow us to debug the problem.

From: Brian Standley
Reply-To: MDSplus/mdsplus
Date: Tuesday, January 12, 2016 at 11:13 AM
To: MDSplus/mdsplus
Subject: Re: [mdsplus] twf-fix-compress-again2 is no good (#299)

I was able to reproduce the segfault using an ACQ196 in my office with a different server running 7.0.206 -- initftp, provide external trigger, store. Reverting back to 7.0.205 fixes it. I switched back and forth between the two versions a few times just to be sure. I can provide the test model later in the evening if it helps.

from mdsplus.

Absolutely!

from mdsplus.

Josh was able to reproduce the problem here so we should be able to debug it locally. No need to send us your tree.

from mdsplus.

Unbelievable.... I made a stupid typo (-1 instead of -i) when I was
trying to make the code more readable! I'm calling it "Ken's revenge".
I think #300 should fix it.

On 1/12/2016 11:56 AM, Tom Fredian wrote:

Josh was able to reproduce the problem here so we should be able to
debug it locally. No need to send us your tree.

—
Reply to this email directly or view it on GitHub
#299 (comment).

from mdsplus.