<fundies>
so can you patch capnproto to work on windows
frigginglorious has quit [Ping timeout: 255 seconds]
<kentonv>
fundies, sorry, but you need to solve the problem in your code. There's nothing we can do in Cap'n Proto.
<kentonv>
many other people have managed to work around this problem
<fundies>
none should have to tho
<fundies>
i have no issue with other libraries
<kentonv>
yes, well, unfortunately Cap'n Proto chose to use VOID as a (namespaced) constant name, long before anyone tried to use it on Windows. We can't just change the name, it will break lots of existing code.
<fundies>
why not they can use macros to adhere to the old broken spec instead
<kentonv>
I don't know what you mean
<fundies>
they can define VOID notvoid
<kentonv>
who can?
<fundies>
ppl who dont want to fix their code to deal with you fixing your mistake
<kentonv>
that's not how software engineering works, fundies. Already-working code takes priority over code that hasn't been written yet.
<simpson>
fundies: You're asking for an API change in one of the oldest and most important schemata. You'd be asking every capnpc plugin author, like myself, to update my code, even if I don't use C/C++.
<kentonv>
existing working code shouldn't break when you upgrade Cap'n Proto to a new version.
<fundies>
if you made a fundamental mistake in design it should
<simpson>
This is a flaw either in the Win32 source-code-level API or the C language, depending on POV. It sounds like there's a reliable workaround.
<kentonv>
with all due respect, fundies, it sounds like you haven't maintained a widely-used library before
<fundies>
kentonv, with all due respect I havent seen any library make this mistake
<kentonv>
fundies, I'm not interested in continuing this conversation. Please just don't use Cap'n Proto.
<fundies>
kk i'm sure flatbuffers or something else actually works
<fundies>
without any workarounds needed
harish has joined #sandstorm
<simpson>
kentonv: Hm, is there a canonical name for the parser/reader/unpacker/loader/etc.? The thing which is dual to the builder?
<kentonv>
simpson, usually "reader"
<simpson>
Okay, sweet.
ogres has quit [Quit: Connection closed for inactivity]
pie_ has quit [Remote host closed the connection]
pie_ has joined #sandstorm
pie__ has joined #sandstorm
pie_ has quit [Read error: Connection reset by peer]
harish has quit [Remote host closed the connection]
pie__ has quit [Ping timeout: 248 seconds]
harish has joined #sandstorm
frigginglorious has joined #sandstorm
frigginglorious has quit [Ping timeout: 248 seconds]
frigginglorious has joined #sandstorm
frigginglorious has quit [Ping timeout: 260 seconds]
jemc has joined #sandstorm
jemc has quit [Ping timeout: 255 seconds]
harish has quit [Ping timeout: 240 seconds]
<ocdtrekkie>
Having issues uploading a file to Davros on Oasis.
<ocdtrekkie>
Didn't have any issues as recently as yesterday.
<ocdtrekkie>
Tried three different browsers.
<ocdtrekkie>
Any ideas?
frigginglorious has joined #sandstorm
frigginglorious has quit [Ping timeout: 240 seconds]
Telesight has joined #sandstorm
Lionel_Debroux has joined #sandstorm
Lionel_Debroux_ has quit [Ping timeout: 240 seconds]
logicfish has joined #sandstorm
frigginglorious has joined #sandstorm
frigginglorious has quit [Ping timeout: 248 seconds]
pie__ has joined #sandstorm
harish has joined #sandstorm
Isla_de_Muerte has joined #sandstorm
NwS has quit [Ping timeout: 268 seconds]
mtgco has joined #sandstorm
jemc has joined #sandstorm
frigginglorious has joined #sandstorm
frigginglorious has quit [Ping timeout: 258 seconds]
pie__ has quit [Ping timeout: 240 seconds]
jemc has quit [Ping timeout: 240 seconds]
Lionel_Debroux has quit [Ping timeout: 240 seconds]
logicfish has quit [Quit: leaving]
frigginglorious has joined #sandstorm
frigginglorious has quit [Ping timeout: 268 seconds]
<kentonv>
xet7, it looks like I need to do a bugfix release, then you should be able to upload.
<xet7>
Ok :)
<kentonv>
it appears that under node 8 (which meteor recently switched to), assigning to an undeclared variable no longer puts it on the global object. Which is great! But it turns out that various bugs that were previously masked by global handling are now throwing exceptions.
<kentonv>
I did run all the tests but our coverage just isn't good enough. :/
<kentonv>
separately, it appears node now aborts the process when an HTTP client decides to disconnect before they've received the full response. WTF?
<kentonv>
I must be missing something because that's a ridiculously stupid behavior, if that's what it's doing.
<kentonv>
JFC, it also aborts when it fails an SSL handshake
Lionel_Debroux has joined #sandstorm
Lionel_Debroux has quit [Client Quit]
Lionel_Debroux has joined #sandstorm
<TimMc>
kentonv: When you say "aborts the process", is that just the process or thread handling that one connection, or the whole app?
<kentonv>
the whole node.js process -- this is for the sandstorm shell UI, not for an app
<kentonv>
I'm seeing node.js reporting "socket hang up" as an uncaught exception
<kentonv>
node aborts on uncaught exceptions
<kentonv>
but I have no control over the HTTP handling code where this exception is being thrown
<ocdtrekkie>
Heh, just saw xet7's email and was like "that sounds familiar to what I saw last night".
<xet7>
:D
<kentonv>
ocdtrekkie, did you see the same exception message somewhere?
<ocdtrekkie>
Think I saw something like it in the debug log.
<ocdtrekkie>
Though the particular point was that his log talked about the destructor on an upload.
<ocdtrekkie>
And my uploads get stuck at the end.
<ocdtrekkie>
Like it doesn't know what to do when it's finished getting the data.
<ocdtrekkie>
It's not throwing it for me right now, but this is in the debug log.
<ocdtrekkie>
events.js:160
<ocdtrekkie>
throw er; // Unhandled 'error' event
<ocdtrekkie>
^
<ocdtrekkie>
at IncomingMessage.onReqAborted (/opt/app/node_modules/multiparty/index.js:183:17)
<ocdtrekkie>
Error: Request aborted
<ocdtrekkie>
at emitNone (events.js:86:13)
<ocdtrekkie>
at IncomingMessage.emit (events.js:185:7)
<ocdtrekkie>
at abortIncoming (_http_server.js:280:11)
<ocdtrekkie>
at Socket.serverSocketCloseListener (_http_server.js:293:5)
<ocdtrekkie>
at emitOne (events.js:101:20)
<ocdtrekkie>
at Socket.emit (events.js:188:7)
<ocdtrekkie>
at TCP._handle.close [as _onclose] (net.js:492:12)
<ocdtrekkie>
Though that might be from when I gave up and closed the app.
<kentonv>
I think it may also be that frontends are dying all the damned time because of this node problem, so if you upload something large enough then you're sure to have the connection aborted
<kentonv>
currently uploading fix, should be live in like 60 seconds
<kentonv>
well 90 seconds
<ocdtrekkie>
No prob, I'm still waking up.
<ocdtrekkie>
Just saw you asked about what I saw, didn't want to fail to give you info.
<kentonv>
ok guys, I have to run out for an hour, but I deployed a couple fixes to Oasis. ocdtrekkie, can you check if it's better? xet7, unfortunately spk uploads go to Alpha which isn't patched yet, I'll get to that when I'm back.
<ocdtrekkie>
Will take a look.
<ocdtrekkie>
Oasis still claims it's on 0.217, despite having a 0.219 changelog, FYI.
<ocdtrekkie>
Upload works great though!
<xet7>
kentonv: spk publish to Alpha has same error, I'll wait for you to fix it. Meanwhile I'll build VirtualBox image of Wekan release.
isd has joined #sandstorm
<ocdtrekkie>
Yeah, he hasn't updated Alpha yet which is where the app index is.
<kentonv>
ocdtrekkie, the about page on Oasis lies when I do a hotfix
<kentonv>
hotfix = "update the Meteor app without updating the whole system", which I can do without incurring downtime
<kentonv>
the real thing to pay attention to is the changelog, which will show the actual latest version at the top
<kentonv>
though I didn't update the changelog before pushing this fix
<ocdtrekkie>
0.219 wasn't a hotfix though, right?
<ocdtrekkie>
Or it was because you were just updating Meteor stuff.
<kentonv>
yeah I often do "hotfixes" for normal Oasis updates because a full update requires downtime, and if only Meteor stuff changed then it's pointless
<kentonv>
OK on further inspection I think it's actually outgoing HTTP connection failures that cause the crashes, i.e. probably mostly from TinyTinyRSS
<ocdtrekkie>
That seems plausible, TTRSS creates a lot of outgoing connections.
<kentonv>
I guess there's probably some way that I could be catching these exceptions gracefully
<kentonv>
(whereas with incoming, I don't think there's anything I could do)
Telesight has quit [Remote host closed the connection]
<ocdtrekkie>
xet7: I'll review it soon. Looks like there's a decent handful of changes to test out. I saw you closed the emoji issue that's been out there a long time.
ogres has quit [Quit: Connection closed for inactivity]
<xet7>
You can use markdown to include any image in card, checklist etc. Even big images :D
<xet7>
It makes very fun looking cards
<xet7>
and having big font sizes with adding # in front of text
<xet7>
So I'm thinking have we lost anything
<xet7>
Trello does not have markdown in cards :D
<xet7>
We also have attachment images as slideshow :D
<ocdtrekkie>
Does normal Unicode emoji show up in Wekan?
<ocdtrekkie>
It does. :D
<xet7>
Nice :)
<ocdtrekkie>
So the only thing really missing is like something that autoconverts :D to 😀
<ocdtrekkie>
And a lot of modern OSes have proper ways to insert emoji directly. Mobile keyboards especially.
<TimMc>
ocdtrekkie: ugh please no autoconversion
<ocdtrekkie>
I think Wekan used to do it, and now it's broken. I'm kinda in favor of letting people specify it themselves these days.
<ocdtrekkie>
As noted, a lot of OSes have a way for people to insert emojis if they want them, and it's really a holdover from the old school days when web apps served images over symbols representing emojis.
<xet7>
Autoconversion to emoji did break inserting MAC addresses.
<xet7>
so fixing it broke autoconversion :D
<kentonv>
+1 to no autoconversion, or if you must have it, do it on write, not on read, and let the author opt out
<kentonv>
the most infuriating thing is when there's an option to disable emojis, but it only makes me not see emojis, everyone else still sees them in my messages
<kentonv>
I don't care if other people want to put emoji in their messages, I just don't want them in *my* messages. :P
<ocdtrekkie>
Well, good news, everyone, because 0.55 works as everyone here wants it to. :D
<TimMc>
++
<isd>
I love it when pidgin inserts emoji in the middle of ipv6 addresses.