"Solr makes extensive use of MMapDirectory, which uses RAM not reserved for the JVM for most of the Lucene index. Therefore, as much memory as possible should be left for the operating system to use for this purpose. The heap allocated should be as small as possible while maintaining good performance. 8-16Gb is quite common..."
d4rk joined the channel
yvanzo[m]
atj: My bad, rsync has to be installed too.
atj[m]
AFAIU, increasing the Solr heapsize will just lead to more GC time and probably reduce performance. We need enough spare RAM to fit the indexes in the page cache.
yvanzo[m]
reosarevok: I would avoid repeating the type name to start any sentence.
atj[m]
I would expect to see disk I/O to drop significantly in that case
yvanzo[m]
reosarevok: _Usually vertical_ is a quite commonly distinctive feature of posters, and it might help a bit with distinguishing it from banner.
atj[m]
yvanzo[m]: rsync and rrsync, right?
zas[m]
atj: yes, since we can easily change the settings on one node, it would be an interesting experiment to see how it compares to others. If it doesn't have positive changes, we just revert.
reosarevok[m]
grrlsync
d4rk has quit
d4rk joined the channel
yvanzo[m]
atj: You’re right, see the script `install_rrsync.sh`
atj[m]
zas[m]: i have no issue with tweaking some stuff to see what impact it has. this would be slightly more involved than just changing the `-Xms` and `-Xmx` values as I've configured Solr to use hugepages.
nb nb I was just looking over your commits and running the updated code
Jade[m]
Nice!
Just pushed my MVP passing data to the templates
It's not perfect yet, need to add some more error handling and add a way to pass on the preview routes
bitmap[m]
how was your meeting with aerozol? looks like you've already implemented the mockups he made, or is it still a WIP?
Jade[m]
bitmap[m]: It went well, he's been creating some designs in figma
I've implemented the subscription one
I'm not sure I've got the font scale right, 12px is a bit small imo
but otherwise looking godd
s/godd/good/
bitmap[m]
Cool! Hmm, is there a way to preview the template response HTML from the swagger UI that I'm missing?
Jade[m]
No, unfortunately
Swagger seems to be primarily focused on JSON
I've been using the /send route with Mailpit since I added the parameters
bitmap[m]
Np, it's trivial to navigate to get to it manually anyway
Do you know why all of these <!--[if mso | IE]> are necessary? Is Outlook really not able to render tables correctly? 😂
Jade[m]
Yeah
Honestly not really, it's just what MJML is adding by default
A lot of older outlook versions would be outdated IE though
And we all know how bad IE was
bitmap[m]
Yeah, I expect the newer versions would be using Chrome, but I wonder how old these versions of Outlook are and whether we need to support them (if it's configurable)
Since it seems to add quite a bit to the generated email size
Jade[m]
Yeah, possibly something to look up
Although it doesn't matter too much
Jade[m] uploaded an image: (53KiB) < https://matrix.chatbrainz.org/_matrix/media/v3/download/matrix.org/GUQvxUrHspWpnhevwKyYVtDa/image.png >
bitmap[m]
Looks very clean!
Jade[m]
Thanks! :)
reosarevok[m]
Keep in mind a space is missing there before the parens :) Might be known
bitmap[m]
You said you can pass parameters now -- I assume that's what you've done with the name, do you have an example request for that?
Jade[m]
reosarevok[m]: Oh yeah, that's a simple fix hopefully, just an issue from not escaping properly
bitmap[m]: Yep!
Jade[m] sent a code block: https://matrix.chatbrainz.org/_matrix/media/v3/download/chatbrainz.org/lvUszNAOjmQVSYMTVZpwljdV
yvanzo[m]
<bitmap[m]> "Do you know why all of these <!-..." <- That’s is really the main point of MJML: bringing support for most mail clients.
yvanzo[m]: Sure, but it is a significant portion of the email body, and I'm curious how many of our users actually use older versions of Outlook
d4rk has quit
d4rk joined the channel
Perhaps it also helps with other clients though; I don't think it's that big a deal
Jade[m]
I ended up writing a new feature for mrml this week, https://github.com/jdrouet/mrml/pull/437, which means I can pass around groups of Mjml elements and still have then behave correctly
yvanzo[m]
bitmap: Off-topic. It reminds me that we should revisit supporter browsers for the MB website too.
Jade[m]
Jade[m]: Once that's merged and published, along with the other patches, I should be back to using upstream realeases
d4rk has quit
yvanzo[m]
s/supporter/supported/
d4rk joined the channel
bitmap[m]
Very nice. That seems like a pretty essential feature to have
Jade[m]
Haha yeah, at least for the way I'm using it
Also added that redirect you asked about last week ;p
bitmap[m]
I noticed, thank you :)
d4rk has quit
d4rk joined the channel
So with work on the templates already progression faster than expected, will you be looking into translations soon?
s/progression/progressing/
Jade[m]
Yeah! I'm going to start working on that as I implement the templates more properly
I've got a plan and I've got code from last summer, I'm hoping it goes well 😅
I think I'm going to try use the MessageFormat syntax
bitmap[m]
You decided against Fluent?
Or does Fluent support MessageFormat syntax?
Jade[m]
Yeah
I generally preferred the MessageFormat syntax, and it works better with IDE extensions
d4rk has quit
Plus Fluent works much more at runtime, whilst MessageFormat has quite a few libraries that let you parse all the translations at compile time
d4rk joined the channel
If that's OK?
yvanzo[m]
Both are OK 👍️
(Or either?)
bitmap[m]
Yeah, I'm fine with using MessageFormat start with. There are few enough templates that I don't think it would be a significant problem to use either if we run into problems with the other
Jade[m]
OK great!
yvanzo[m]
MessageFormat v2 is more recent and backed by a large community (Unicode).
bitmap[m]
I'm more familiar with Fluent at this point, so will have to read up on MessageFormat v2 a bit more. I assume you're using v2 then?
Jade[m]
So I think I'm using the v1 syntax
yvanzo[m]
Fluent v1 has Weblate support and implementations in Rust, Python, and JS too, and a v2 version is in progress but probably not ready any soon.
Jade[m]
It's all a bit confusing, cause the main MF library has version 2.0 that uses the MF1 syntax
Plus everything about MF2 is in beta
so it's the syntax that looks like 'Today is {T, date}',
I think the MF2 spec is still a draft, last I checked?
yellowhatpro[m] has quit
But it's the one they want to eventually add to the browser Intl API
yvanzo[m]
“in Tech Preview”
Jade[m]
IIRC they want to have a translator that turns MF1 -> MF2 by the time they're fully ready, too
bitmap[m]
Perhaps we can move when the MF2 ecosystem is more stable then
Jade[m]
Yeah
yvanzo[m]
Jade: Don’ worry to much about it. Pick the one that is the easiest for you to work with at first. So we can see the benefits and drawbacks for a real use case. We can decided to move to another format if needed later on (I mean after GSoC).
+t +o
-d
bitmap[m]
Jade: Anything else you you wanted to show or discuss today?
Sounds like we're all agreed on the starting point for translations
d4rk has quit
d4rk joined the channel
yvanzo[m]
Sorry for the approximate typing, just trying to keep up with the pace 😬
Jade[m]
Nope, I think that's it!
yvanzo[m]: No worries
See you around!
bitmap[m]
Thanks! enjoy your evening
Jade[m]
You too! :)
<Jade[m]> "IIRC they want to have a..." <- Turns out they already have it!