and fixing #599 will unblock pushing out more things?
yvanzo
I worked on moving macros to React too.
Yes, almost every page using localization will be unblocked with #599
ruaok
any idea of how many pages we've done the work for that will be releasable or near releasable one #599 is taken care of?
chhavi_
Okayy :)
adhawkins joined the channel
adhawkins has quit
adhawkins joined the channel
yvanzo
No, there are many macros that need to be replaced as well, but that is a root issue for migrating even macros.
ruaok
ohh, fun. circular dependencies.
I'm trying to figure out how to manage the react move, the UX cleanup and schema change in the next few months.
clearly the react move needs to be done before we can really dive deeper into the UX stuff.
yvanzo
hmmm... no, #599 unlocks l/lp/ln macros which are used pretty much everywhere. Then, we have to migrate other root macros (paginator, and so on), then we have full pages.
Only basic pages can be moved soon, such as instrument list, isrcs, iswcs (by spellew).
ruaok
migrating other root macros... 1 week? 1 month? 6 months?
I'm wondering if moving something like the artist page could also give it a new UX, but I fear that is going to be too disruptive to the flow our our UI.
and doing it in separate passes will be duplication of work.
yvanzo
More than 1 month for sure.
ruaok
ok
chhavi_: a picture is emerging now. :)
yvanzo
For every bit we move to React, we would like to write Selenium tests as well.
ruaok
makes sense.
I'm going to think out loud for a moment....
chhavi_, yvanzo: see what you think of this.
1. we delay the schema change to autumn.
2. While bitmap and yvanzo work to move to react, chhavi_ lines up all the design and componentization.
3. once both the UX prep and the move to react are done, we call "all hands on deck." every engineer works on the UX improvements to MB. we stop doing releases on MB, except for critical bugs.
4. once we finish all non-edit pages we release and resume a release schedule.
this approach serves two purposes:
1. For a while we're all working on the UX stuff. chhavi_ plays master of ceremony for this process. we learn and move MB along.
2. It is less disruptive to non-editors. We'll end up with a multiple personality disorder site, where edit pages are old, and non-edit pages are modern.
The theory is that our editors are more able to cope with #2 than the general public.
depending on schedule we can then see if we focus on an autumn schema change or continue with edit page UX migration.
not sure how realistic this process is. but it is a starting point for us to think about.
chhavi_
Sounds good to me, from my limited knowledge :p :)
ruaok
I think the most important thing for me is that you're not actually doing any of the nitty gritty work. but really more mastermind that moves the whole process along.
we'll need to see how GSoC fits into this, because you do need to write code for it.
first I need to get bitmaps feedback, then we can see.
samj1912
Just 1 thing I would want to do before we switch gears to UX is to scale our postgres and setup our read replica properly.
ruaok
that is becoming important, yes.
maybe you can focus on that while the react train is moving.
and did you see the spreadsheet about hosting costs?
samj1912
Yup i did
ruaok
going full cloud is vastly more expensive than hetzner.
hetzner cloud might be interesting, zas and I will examine.
samj1912
Cool, oh and did you also compare with aws? I thought that might be cheaper. Not sure though.
ruaok
AWS was not directly in there as google cloud was something we could get some support from google on.
but they are priced in similar ways.
certainly not different enough to dislodge hetzner from the logical choice.
Dumbo joined the channel
yvanzo
ruaok: I agree about focusing on requirements for UX stuff. While I dislike delaying schema change once again, I am not sure we can make use of it before the autumn anyway, except if it helps improving UX for non-editors.
ruaok
are there things schema-change wise that are pressing for us to address?
yvanzo
Schema change tickets I reviewed til now are all for editors, e.g. MBS-5387