search

so-90prime / 90prime

General repo for everything related to 90Prime (docs, diagrams, thoughts)
2 stars 0 forks source link

Final elimination of the galilserver/fw-gui forever #48

Open dsand opened 2 years ago

dsand commented 2 years ago

The issue discovered during the Jan 2022 run is that Azcam pulls header keyword information from the old galilserver/fw-gui. So we need to work with @mplesser to replace this functionality in Azcam to talk to the new galil code.

dsand commented 2 years ago

Notes in Feb 3, 2022 agenda from @pndaly

Update on removing galilserver/fw-gui forever: Houston, we have a problem. I have taken a look at this and there are several alternatives each with its pros/cons: a. Use PyIndi as distributed with INDI. This looks OK but we would have to install all of INDI plus the PyIndi SWIG wrapper (so it needs a C compiler) on the Azcam machine which runs Windows; b. Use pyindi as written by Scott/Tim at MMT and used by Dan in the web-GUI. This looks like the best approach but looking at the code, it seems very web-oriented which is not what we want. I think I’ll contact Scott about this to see if he can spare some time on a Zoom call to discuss the code; c. Use GCLIB and produce a Python copy of Galil_DMC_22x0_CLI (which was written in C and works very well). This requires installing GCLIB etc on the Azcam machine under windows. A further drawback is that we would have 2 points of control for 90Prime which we want to avoid; d. Similar to c) is to write an NG-server and implement the commands identified by Dan/CJ/me. This has promise but has the same drawback as we would have 2 points of control); e. Move Azcam off the windows machine to a Linux environment? I think I shall pursue b) further but it won’t be quick so I doubt it’s be available for the Feb run.

pndaly commented 2 years ago

Met with CJ, Scott and Dan after the s/w meeting this morning. We ironed out a path forward and Dan/Scott are happy to help. Both points (b) and (d) were discussed and both have merit so we have decided to follow (b) as Plan A and (d) as Plan B. The reason we need a 'Plan B' is we're not sure how and asynchronous class would integrate into Mike Lesser's code. A follow up meeting will be held to check on progress.

dsand commented 2 years ago

The new NG-server is working well, both on @pndaly 's end and @mplesser 's with Azcam. As of April 21, we did fix a filter related bug, so I will keep this open for another run or two in case we need to iron out further things.