00:29:10 <11Eino Keskitalo> @mumra I wonder if we could leverage GitHub workflows for some e2e tests for the webserver 🤔 00:33:08 <06mumra> would certainly be not too hard to do a basic smoke test of: does the server start up, does it render a page with 200 OK status, does it contain the string "welcome to webtiles"? 00:33:18 <06mumra> however, it wouldn't have prevented this issue 00:33:39 <06mumra> unless also combined with automatic restarting of the server 00:35:34 <06mumra> but yeah could set up a real simple E2E workflow some time have a spare hour or two 00:59:14 -!- indigaz26 is now known as indigaz2 03:17:03 <12asciiphilia> Hmm, my DWEM just broke because my commit didn't recognize static/scripts/contrib/require.js?v=f98d0876e9b80f5499dda72093621588950b9708b4261c8aa55912b7e4851e03596185486afb3a9a075f90f59552bb9ec9d2e67534a7deb9652ba794d6ee188d. Why does require.js need a v hash? 03:17:47 <12asciiphilia> Actually, I haven't been able to get off work yet, so I still don't know what that commit does. 03:22:38 New branch created: pull/4844 (1 commit) 13https://github.com/crawl/crawl/pull/4844 03:22:38 03WizardIke02 07https://github.com/crawl/crawl/pull/4844 * 0.34-a0-996-g058a6bd30b: Don't display short lived frames in console 10(71 minutes ago, 76 files, 49+ 234-) 13https://github.com/crawl/crawl/commit/058a6bd30b10 03:31:44 Experimental (bcrawl) branch on underhound.eu updated to: 0.23-a0-5261-gd9800d219b 03:31:58 <07wizardike> It's to prevent an old version of a file being used from the browser's cache. require.js probably doesn't really need one as it's unlikely to change, but they were added to all files so it got one 03:33:35 <12asciiphilia> Isn't there already a hash in gamedata/*/.js? Is it a little different from that? 03:38:35 <07wizardike> I think there is the game version in there somewhere, but it's the same for multiple versions of trunk for reasons I don't understand 03:40:18 <07wizardike> I think it's just set to "trunk" for all trunk games actually instead of the version 03:47:36 <07wizardike> It might have been better to correct the game version we already have in the url, but I don't understand why it's set the way it is and I also didn't make this commit 03:53:28 03WizardIke02 07* 0.34-a0-1001-g43d75eb032: Don't display short lived frames in console 10(2 hours ago, 76 files, 49+ 234-) 13https://github.com/crawl/crawl/commit/43d75eb032ea 04:10:31 <12asciiphilia> It seems like not only the in-game font but also the lobby font has changed. Is this also an intended behavior? 04:10:38 <12asciiphilia> @mumra 05:30:25 <06mumra> Yeah gamedata doesn't need a hash because it already has the version string, so any scripts in gamedata will never change therefore can never go stale 05:30:47 <06mumra> Yes. Using the same font (as the desktop version) everywhere 05:30:58 <06mumra> But you can override it easily, it's stored in a css var 05:32:26 <06mumra> in CSS css :root { --crt-font: Cool Font, monospace; } in JS js document.documentElement.style.setProperty("--crt-font", "Cool Font, monospace"); 05:34:56 Unstable branch on crawl.akrasiac.org updated to: 0.34-a0-1001-g43d75eb (34) 05:35:38 <06mumra> (for DWM i'm wondering can you check the required module name rather than looking at the URL? as it's still requiring "./client" or whatever, the requirejs config is just configured to add the hash) 05:46:07 <06mumra> %git 43d75eb 05:46:07 <04Cerebot> WizardIke * 0.34-a0-1001-g43d75eb032: Don't display short lived frames in console (4 hours ago, 76 files, 49+ 234-) https://github.com/crawl/crawl/commit/43d75eb032ea 05:46:29 <06mumra> @wizardike is it possible this commit stopped the skills menu showing changes on webtiles? 05:48:13 <06mumra> https://crawl.dcss.io/ is on that latest version now, but I don't see the same issue on CBRO which is a couple of commits earlier (was just trying to confirm it wasn't my changes that somehow caused this!) 05:48:27 <06mumra> CDI going through the wars today 😭 05:49:37 <07wizardike> That's very possible, the old tiles code used the console updating as a heuristic to determine when to redraw the screen. I'll look into it 05:50:14 <06mumra> basically if you toggle skills it doesn't update the screen, but if you exit the menu and come back in you can see the skill was toggled 05:50:24 <06mumra> so yeah seemed likely it was down to rendering 05:52:44 <06mumra> now this is the kind of issue that some decent E2E tests could catch 🙂 06:01:10 <09gammafunk> exactly, we don't have the supporting infrastructure or guarantees. Perhaps someday we can work towards a nice standard containerized model like that, but certainly it's not viable today 06:08:05 <06mumra> I mean containers also work best in a situation where you can scale horizontally by spinning up more containers ... would be very difficult to make crawl webtiles work like that at all and a single beefy server just gives the performance you need 06:26:22 03WizardIke02 07* 0.34-a0-1002-g7d1d168b82: Fix the skills menu not redrawing in webtiles 10(12 minutes ago, 1 file, 4+ 0-) 13https://github.com/crawl/crawl/commit/7d1d168b820e 06:41:04 <09gammafunk> it's not about scaling, it's more about management 06:41:46 <09gammafunk> this is just an old, hand-rolled admin system that dates from a long time ago when e.g. containers didn't exist, kernels had less functionality, etc 06:42:30 <09gammafunk> cnc uses a container, for example, and it'd probably be good to modernize our default live server infrastructure along those lines 06:43:03 <09gammafunk> not proposing we move away from a single server model, just saying that we could make these sorts of issues easier for admins 06:43:39 <09gammafunk> this is a wishlist item anyhow, not realistic that we actually get it any time soon, what with how old some of our servers are 07:10:58 -!- rozlav82 is now known as rozlav 08:47:17 <09hellmonk> minor bug: rending blade's mp-dependent power calls const int mp_spent = max(0, you.magic_points - spell_difficulty(SPELL_RENDING_BLADE)); to set the amount of mp contained in the blade. If something modifies the casting cost of the spell, this value will be incorrect. 10:16:00 <06mumra> @wizardike Unfortunately the rendering changes seem to have affected Faded Altars as well 10:16:57 <06mumra> When you pray at one now you don't see the message about which gods it is, until you press Esc then it dumps the whole message 10:17:09 <06mumra> Hmm, offline unfortunately 10:17:14 <06mumra> This is a bug on CDI right now 10:20:40 <06mumra> And having seen a couple of side effects now I can't help but wonder if there's even more 10:24:29 Unstable branch on crawl.akrasiac.org updated to: 0.34-a0-1002-g7d1d168b82 (34) 10:25:44 <06mumra> Found another problem, pressing & you don't see the wizmode prompt 10:27:46 <06mumra> I'm assuming these issues affect console as well 10:35:08 <06mumra> Wondering if we should revert 43d75eb and 7d1d168 for now and let WizardIke have another look when they're back online ... could be a whole boatload more side effects from this 10:36:06 <06mumra> Hmm interesting, it doesn't affect console. Just webtiles 10:37:35 <03implojin> i'd be in favor of reverting for now if it has a bunch of side effects that weren't realized 10:37:48 <03implojin> can always un-revert later after it's been looked at 10:40:54 <03implojin> feels to me like the concept of only rendering one frame per player turn would have issues any time we're trying to do animations, but i haven't reviewed precisely what this commit touched 10:42:53 <06mumra> It touched a lot of stuff (76 files) 10:55:03 <06mumra> Shatter animation seems fine 10:55:18 <06mumra> It's very obscure which things were impacted by this 11:03:08 <04dracoomega> Probably the most thorough way is to look at every removed redraw call and test wherever it was (though that's obviously slow work) 11:04:14 <04dracoomega> I admit I was a little trepidatious when I looked at the commit, but also there is a fair bit of actually-redundant rendering code in places. (A few such calls came up in my movement refactoring the other day and I spent quite a while testing things in webtiles and local tiles to convince myself that it really actually wasn't doing anything and could be removed - though I might still be wrong! >.>) 11:05:07 also remember that some redraws are redundant with respect to display, but are needed to provide syncpoints for ttyrecs 11:05:58 <04dracoomega> Oh? Howso? 11:11:03 <06mumra> With somewhere approaching 200 removed calls across 76 files yeah that's quite a big task 🙂 (and of course with the nature of crawl's code, many of those places could be reached by any number of routes that may or may not be affected differently by the removal) 11:11:18 <04dracoomega> Yeah 11:13:18 <04dracoomega> (Some of the slowness in what I've been doing lately is because I've been trying to trace a large number of things back to every place that could call them, which is sometimes pretty large.) 11:13:35 <04dracoomega> Like, "Do the differences between these two methods here really matter?" sort of things 11:14:50 <06mumra> "circuitous" is not even an adequate word sometimes 😉 12:20:45 <09gammafunk> crawlcode is an infinite loop... 12:40:00 it's a pile of spaghetti 12:41:36 -!- jds_ is now known as jds 12:41:43 I love spaghetti 13:07:38 03WizardIke02 07* 0.34-a0-1003-g34f3ec3571: Fix webtiles not redrawing when showing a prompt 10(10 minutes ago, 1 file, 3+ 0-) 13https://github.com/crawl/crawl/commit/34f3ec3571aa 13:32:38 <07wizardike> It uses the exact some logic as tiles to decide when to show a frame. So anything that calls delay will show a frame which animations have to do as they wouldn't show up the majority of the time anyway without it with the old logic 14:30:32 -!- HeniMiku is now known as vagabondtruffle 15:41:19 Unstable branch on underhound.eu updated to: 0.34-a0-1003-g34f3ec3571 (34) 16:19:29 -!- Myth is now known as Autumn 16:30:35 <12asciiphilia> Since the module information is extracted from requirejs, the module name can be recognized even if it has a hash. but the CNC's trunk patcher that replaces the initial requirejs script was designed to replace the script tag with a regular expression, so it was not loading. 17:26:08 <02Darby> end of an era of chronomancy 17:26:12 03DracoOmega02 07* 0.34-a0-1004-g5f9b90cd0b: Fix being able to freeze time on other floors by peeking at them (Darby) 10(2 minutes ago, 1 file, 5+ 4-) 13https://github.com/crawl/crawl/commit/5f9b90cd0b5c 17:28:29 <04dracoomega> Pushing the fix for this particular discovery on my long journey into finding more and more broken things straight to master >.> 17:29:16 <04dracoomega> I can't even count how many levels deep I am now 17:29:48 <04dracoomega> Something I was using to verify a fix to a different issue today turned out to itself be buggy and need fixing 17:46:58 <03implojin> oh wow good bug 22:35:28 Unstable branch on crawl.develz.org updated to: 0.34-a0-1004-g5f9b90cd0b (34) 22:58:36 Windows builds of master branch on crawl.develz.org updated to: 0.34-a0-1004-g5f9b90cd0b 23:35:33 Unstable branch on cbro.berotato.org updated to: 0.34-a0-1004-g5f9b90cd0b (34) 23:48:50 <06mumra> Ohhh got it 23:55:36 Monster database of master branch on crawl.develz.org updated to: 0.34-a0-1004-g5f9b90cd0b