“Not V2492Cyg again!” … is what the regular reader of these posts might shout. I know! But this one is special, and a follow up on the post from week 316 in January. The plot is a screenshot from a copy of the HOYS website, which interacts with the newly structured database. This is the first successful query with the altered web interface – there have been many unsuccessful ones over the last two weeks 🙂
The entire search software had to be re-written to work with the new way the data and tables are laid out. As everything is cross-matched with Gaia already, this is far quicker now. The new version contains about 3 times as much data as the currently live database, but the query is completed in just under a second, where most of the time is spend organising the output so it can be nicely displayed on a webpage. The query of the much smaller amount of data in the old version is taking longer than it takes to write these lines ….
It sounds like an easy job to rewrite the code, but it is far from it. I learned a lot (mostly because I did not write the original software). How do you set up a working copy of the interactive website so it can be tested on the same machine without interfering with the existing setup? How do you set permissions for users on databases in MySQL? How do you transfer hidden attributes for files (cp does not do that)? How do you query database tables with Django and set up dynamic classes/models for that? And most importantly: How do you best argue with ChatGPT to make it rewrite your code so it does what you need, creates the right output, is as fast as possible, and is in the correct version? These are just a few of the things ….
Now: With this step done, we will slowly have to work our way through a list of all the tasks that the website performs (adding, editing, deleting data, etc.) which are different in the new structure. All these functions have to be adapted, and rigorously tested. Then the data in the live version that is not yet in the new structure needs to be transferred and the automatic cross matching with Gaia implemented before we can put this version live. This will include a look at improving the speed of the data processing (in particular the photometry calibration). We will keep you updated on the progress.