qucsconv is not working under Linux

[Kathy626]

Hey, all I have met a problem relates to qucsconv in Linux. I've tried to use qucscpnv to convert a touchstone file (s38p) into qucsdata file (dat), this command works totally fine in my Windows machine, but it errored out in Linux, and I have attached error message in here: Does anybody has met this problem before?

[Kathy626]

I've found out that since the touchstone file I am using is s38p file, which for each frequency point is 38 x 38 matrix. Not sure whether Linux qucs has the ability to handle this big file

[in3otd]

these errors come from the routine that parses the S-parameters file; it may be somehow related to #789. I'm currently doing some simulations with a 36-port S-parameters file and it works fine in Linux - could be faster probably but the results are ok. The problem might be related to the line terminations, you could try to change that using dos2unix or similar and see if it makes a difference. It will of course be useful if you could share a stripped-down/reduced version of the problematic file.

[Kathy626]

thanks a lot, very very helpful

2018-03-27 12:11 GMT-07:00 in3otd notifications@github.com:

these errors come from the routine that parses the S-parameters file; it may be somehow related to #789 https://github.com/Qucs/qucs/issues/789. I'm currently doing some simulations with a 36-port S-parameters file and it works fine in Linux - could be faster probably but the results are ok. The problem might be related to the line terminations, you could try to change that using dos2unix or similar and see if it makes a difference. It will of course be useful if you could share a stripped-down/reduced version of the problematic file.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Qucs/qucs/issues/790#issuecomment-376641176, or mute the thread https://github.com/notifications/unsubscribe-auth/Aj7O38R3THMA-Y3Wp0HehgFIg-dJq8c9ks5tio79gaJpZM4S37xA .

[Kathy626]

Which file you'd like me to share? the reduced touchstone file or compiled code?

[in3otd]

I'd like to have a Touchstone file which you are not able to convert on Linux so I can try to use that to debug this issue locally.

[Kathy626]

The file is too big to upload, even after I zip the file still too big. The touchstone file I have is 38 ports (s38p), with 1315 frequency sweep points.

[Kathy626]

​ Test_sparams_38ports.zip https://drive.google.com/file/d/1Z09_rcd2lg_VXzbbrOINgkgxuYM_5v9g/view?usp=drive_web ​does this work?

2018-03-27 13:30 GMT-07:00 in3otd notifications@github.com:

I'd like to have a Touchstone file which you are not able to convert on Linux so I can try to use that to debug this issue locally.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Qucs/qucs/issues/790#issuecomment-376664571, or mute the thread https://github.com/notifications/unsubscribe-auth/Aj7O31JrUU414dvGoBgL-JplQBQdBogKks5tiqFWgaJpZM4S37xA .

[in3otd]

thanks for the file; here I'm able to convert it without errors both in Linux and Windows. I've run a quick simulation, the results seem a bit strange (magnitude often >> 1 and not smooth) but I guess this may be ok. Not sure where the problem could be there; you may try to reduce your file to just 1 or 2 frequency points and see if it works fine.

[Kathy626]

that's very very strange, can I ask which Bison version you are on

2018-03-29 10:16 GMT-07:00 in3otd notifications@github.com:

thanks for the file; here I'm able to convert it without errors both in Linux and Windows. I've run a quick simulation, the results seem a bit strange (magnitude often >> 1 and not smooth) but I guess this may be ok. Not sure where the problem could be there; you may try to reduce your file to just 1 or 2 frequency points and see if it works fine.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Qucs/qucs/issues/790#issuecomment-377307455, or mute the thread https://github.com/notifications/unsubscribe-auth/Aj7O30uk-5-4qu-U36yq3ETKTi231ykgks5tjRbugaJpZM4S37xA .

[in3otd]

yep, that's strange indeed. I saw errors similar to what you reported only recently when playing with the Bison grammar file to fix #789 but never with the default grammar file. Here in Linux I'm using bison (GNU Bison) 3.0.4 with flex 2.6.0.

[Kathy626]

Ok, maybe the version is the reason causing this strange thing. Because, from my downloaded QUCS Linux package, Bison is 2.7, flex is 2.5.37

But I am so confused of how you could observe the similar error when you compiling QUCS with Bison Grammar, this should not affect

2018-03-29 12:15 GMT-07:00 in3otd notifications@github.com:

yep, that's strange indeed. I saw errors similar to what you reported only recently when playing with the Bison grammar file to fix #789 https://github.com/Qucs/qucs/issues/789 but never with the default grammar file. Here in Linux I'm using bison (GNU Bison) 3.0.4 with flex 2.6.0.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/Qucs/qucs/issues/790#issuecomment-377342740, or mute the thread https://github.com/notifications/unsubscribe-auth/Aj7O30dx_61LLSYCT86N_MONTSjxsa-oks5tjTLsgaJpZM4S37xA .

[in3otd]

what I meant to say is just that I know these errors come from the parser not recognizing some parts of the S-parameters file since while trying to fix the other issue I incorrectly modified the parsing code (the grammar file) and saw similar errors. As said, the parsing code in the current branch/release (which has not been modified since years) appear to work well here.

[Kathy626]

Thanks. By the way, do you know how to change Bison version for qucs?

[in3otd]

bison is using only when compiling Qucs to build the parser source code; if you update your Bison version you need to recompile Qucs to see if the new version works differently. How to update Bison depends on the Linux distribution.