BEAST-Fitting / beast

Bayesian Extinction And Stellar Tool
http://beast.readthedocs.io
22 stars 34 forks source link

How do we want to manage the BEAST library in the future #778

Open galaxyumi opened 1 year ago

galaxyumi commented 1 year ago

Back in March 2022, we noticed that filter curves in the BEAST library are not the total filter throughputs and thus updated the filters.hd5 properly. Soon after we also updated the vega.hd5 to accordingly reflect the update in the filters.hd5. However, we are aware of possibilities that users could easily miss these updates for a couple of reasons.

(1) Currently, there are two ways of getting the BEAST libraries. One is to get files via "beast get_libfiles". The other is to get files manually from Box. Unfortunately, this means that @karllark (at the moment) needs to update files in two different places. To reduce any potential confusion and management effort, I think it would be better to keep just one way of getting the BEAST libraries. What do people think? @cmurray-astro @christinawlindberg @petiay? If you agree, which way do you prefer to keep?

(2) What would be a good way to let the users know which library version they are using? One of Karl's idea is to add version numbers to the vega (and other files). And then hardcode the up-to-date version number into the beast and have it give an error when old library files are used.

(3) Finally, I think we need to work on a new PR to combine all the functions to simultaneously create and update the filters.hd5 AND vega.hd5 in ONE step. Thus, the users (including ourselves) don't need to worry about filter curve updates in the two files separately in the future.

Any suggestions are very welcome!