Doubled pxp file size and slow opening speed when having user-written panel is open

Have you also looked into the Packages folder where panels and such usually store their meta-data? I can imagine that the open panels make backup copies, which seems like your data has suddenly doubled. What happens when you just load the data or close the panels? There was also a script to calculate folder sizes somewhere, but I can't find it now...

Hi chozo, thank you but yes, I checked total wave size by entering below commands so that the Data browser checks all the waves sizes. It's not panels making backup waves, that's why I guess it's something behind the user GUI that happens when saving Igor.

Modifybrowser expandall

Modifybrowser collapseall

Then it is likely that the analysis package is saving some temporary data.  Maybe you search for BeforeExperimentSaveHook() in the code to see if something comes up. But usually temporary data should be visible in some folder. The long startup time could also mean that rebuilding the panels triggers some lengthy calculation somewhere. Usually I would not think that opening even an 1 GB file takes very long. I am afraid that it is difficult to identify the cause without having the panel code (and possibly an example file) at hand.

@jurumy7 How would you feel about sending a copy of the experiment to WaveMetrics support? We never make anything public from support without asking first. If you want us to delete the file after we examine it, we can do that, also.

Thank you @chozo and @johnweeks! I'll try sending this pxp file with all procedures adopted as you suggested!

jurumy7 did send us a copy of his experiment file, and I tracked down the missing Gigabyte. The experiment file has two wave reference waves each of which has references to over 356000 waves, each of 241 points. The referenced waves are free waves, so most methods for figuring out how much data is represented by the waves present in an experiment file didn't find those waves. It is also a fact that an unpacked experiment file doesn't save free waves, but packed and HDF5 experiment files do. The slow save and open of the file is directly attributable to the overhead of dealing with those 2*356000 waves.

I thought y'all would be interested 😊🤯

I have to say that I'm pretty impressed that Igor survived that assault!

Thanks to John's advice, I could find the reason for the free waves. They were for 'wave reference multithread', but the original coder forgot to kill the free-waves-referencing wave after multithreading as written in the Igor Pro manual, which led to the situation john wrote above. After adding few lines to properly remove that wave, files now open/save instantly! Thank you so much, @johnweeks!

Glad I could help!

If those waves aren't needed later, it would perhaps simplify the code to make the big wave reference wave a free wave. Then it would be automatically destroyed when the function is finished executing.