Although I was busy in uni stuff in the last fornight, I managed to get some work done in Download from fido search result implemented.
The ultimate objective was to move Database.fetch over to using Fido. Oh, and to fix everything that broke, after a horrible rebase.
So now, that PR is under review, and should get merged soon.
Somebody (prolly me) will have to migrate the now website to the new theme pretty soon. Also I would get the Quick Examples for the database package PR merged as soonas I get time.
With this, the GSoC period officially draws to a close, and what has been (and will continue to be) a wonderful journey.
Eh so slight change in plans.
DISCOVR implementation has been delayed, and I’ll do it in December. Which means 2 things :
- I’m not gonna forget about it. It is officially in my December todo list.
- I’m potentially gonna ruin people’s holidays.
Hence during this period, I mostly worked on getting 2 of my PRs merged.
BTW SunPy released 0.8 during this time.
These are the not-so-small PRs which got merged :
So I now enter the final fortnight of GSoC, with a lot of documentation to do for the webapp, and the mighty Download from fido search result implemented PR awaits.
So the next fortnight ends. Lots of work done and progress made. I digressed from the proposal and completed (almost 😛) what was supposed to be done after the 2nd evaluation. I completed the website with all cronjobs and all clients working. Imaging and timeseries data both work, with timeseries having interactive plots, thanks to ehsteve.
After a lot of head-banging and PR reviewing, and responding to reviews, SRS Reader and VSO query block level caching are finally ready to be merged. I must admit, that Fido indexing PR is really useful for people. 😀
So officially, the only thing that remains is implementing a client for DISCOVR, and displaying its timeseries data on the website.
There is always this guy at every SunPy meet 😛
Well, this has been a busy fortnight to be honest. I got caught up with other stuff, and hence I need to play catch up.
So I fixed the SRS (now my to-do list has all boxes checked 🙂 ).
Persistent storage in the webapp works on my PC. (Note that the previous line is poorly worded.)
Now I’m thinking of making slight changes to plans, and I plan to get the webapp done as quickly as possible, with DSCOVR implementation towards the end of GSoC.
Oh, and btw VSO query block level caching is finally getting merged.
I know. I know. Lets face it. I’m not gonna come up with good headings this year also.
So, I’m back at uni now, and (proper) work begins.
I start by fixing up the issue with the SRS reader, caused by past year’s files being tarred up. Oh yes, that took some fixing. That threaded downloader is tricky to deal with.
Now, to break a break from SRS (still I have to fix the missing columns issue), I start with the website. For starters, I need to take an input date, and serve the magnetogram plot of that day with AR locations marked on it. Simple, right ?
Deployment server hates you.
The website is a Flask app, deployed using dokku. The app works fine on my PC, but on the server, hehe. It’s still not working as it should, as plotting the magnetogram takes > 60 seconds (coz of downloading and stuff) and the timeout issue still persists.
Well, its deployed on now.dev.sunpy.org. Construction in progress 😛
I know. A better heading must be given. 😆
So the Coding Period has begun, aaaaand it’s time for me to head home from uni, after a long time. 🙂 😛
This means there was a lot of travelling, eating and lazing around …….. 😎
Hence, very little work was done in this fortnight.
Very little = Fixing number in SRS files with dates after 2002-06-15 = Adding 10,000 to a row
Ah, so 3 evaluations this time.
So for most part of May I was busy with my research work lol. 😳
Point to be noted : DavidPS pronounces my name correctly (unlike Cadair, and, well many other people).
During the second half of May, I was tinkering with Flask and I learnt (a little bit) about how sample Flask apps work so that I hit the ground running (calm down, its not a disaster).
I also was trying different things in the magnetogram PR as my axis alignment wasn’t always right.
Right, so, as the community bonding period comes to an end, it has been decided (well, at least I think so) that the webapp will be deployed on Heroku.
First I plan to get the magnetogram working on the app, and then build on top of that as time goes on.
Oh, and in case you thought doge memes are useless, take a look at the front page of The Guardian below.