<FromGitter>
<jaydorsey> good afternoon. was wondering if anyone could point me in theright direction for manipulating some parsed YAML. i’m trying to merge some data into an exact parsed yaml file
<FromGitter>
<jaydorsey> so, in ruby i could convert the yaml into a hash, and jam some other data into the YAML
<FromGitter>
<jaydorsey> end goal is to merge 2 YAML files together, but I’m also just looking at general “how can I jam another key/val pair under ‘bar’"
<FromGitter>
<jaydorsey> I have 2 YAML files, both have same top level key, different subkey, and I just wanted to load & merge them together so i could look in 1 place for lookups
<FromGitter>
<Blacksmoke16> is the structure of the data known beforehand?
<FromGitter>
<Blacksmoke16> or static i should say
<FromGitter>
<jaydorsey> sort of? i’m using the yml files from the ruby faker gem to build a crystal version to teach myself some crystal
<FromGitter>
<jaydorsey> the files all have a similar top level key(s), then it deviates pretty wildly across file(s) from there
<FromGitter>
<Blacksmoke16> and you want to merge multiple files into one object essentially?
<FromGitter>
<jaydorsey> yeah, kidn of as an aside. i don’t think I really need to jam them all into one data structure (and probably dont’ want to), i just wanted to see if i could so i could learn some more about how i could do it in crystal
<FromGitter>
<Blacksmoke16> working with `YAML::Any` directly is kinda a pita if you're not doing real simple stuff
<FromGitter>
<jaydorsey> thanks, that’s actually reassuring to hear. i thought it should have been simple but i’m finding it isn't
return0e has joined #crystal-lang
<FromGitter>
<PenguinOwl> heya
<FromGitter>
<PenguinOwl> ive been trying to update the alpine linux package for crystal from 0.31.1 > 0.32.1, but building the binary crawls to a stop while building, usually around the codegen step
hpyc9 has quit [Remote host closed the connection]
hpyc9 has joined #crystal-lang
ur5us has quit [Ping timeout: 260 seconds]
alex has quit [Ping timeout: 260 seconds]
<FromGitter>
<PenguinOwl> is there a approx. minimum amount of memory that building the compiler takes?
ur5us has joined #crystal-lang
<FromGitter>
<Blacksmoke16> how much do you have?
<FromGitter>
<PenguinOwl> on my metal server, 2gb
<FromGitter>
<Blacksmoke16> yea thats prob not enough
<FromGitter>
<PenguinOwl> sounds about right
<FromGitter>
<PenguinOwl> vm time
<FromGitter>
<watzon> Yeah building the compiler takes a good amount of memory. You probably need 2.5-3 GB minimum
<FromGitter>
<PenguinOwl> any idea on the rough amount though?
<FromGitter>
<PenguinOwl> oh k
<FromGitter>
<PenguinOwl> i built on my 8gb machine a while ago
<FromGitter>
<PenguinOwl> just didnt know the rough estimate
<FromGitter>
<PenguinOwl> thanks!
<FromGitter>
<watzon> I'd probably use a machine with 4 just to be safe, but it shouldn't need much more than that
<FromGitter>
<PenguinOwl> k great
dwdv has quit [Ping timeout: 265 seconds]
<FromGitter>
<randiaz95> hey guys, is there a performance loss if I were to buffer a file reader to line by line instead of reading the entire file to memory?
<FromGitter>
<Blacksmoke16> could benchmark it
<FromGitter>
<randiaz95> ok I see a couple of methods:``` ⏎ ⏎ Instance methods inherited from module IO::Buffered ⏎ buffer_size, buffer_size=, close, flush, peek, read, read_buffering=, read_buffering?, rewind, sync=, sync?, unbuffered_close, unbuffered_flush, unbuffered_read, unbuffered_rewind, unbuffered_write, write ⏎ ... [https://gitter.im/crystal-lang/crystal?at=5e0d6f14833c373f4d9582eb]
ur5us has quit [Ping timeout: 260 seconds]
<FromGitter>
<stnluu_twitter> @randiaz95 if whatever you are using is extending "Buffered", I suspect it does not matter
<FromGitter>
<stnluu_twitter> what likely happens is, the implementation reads a set amount every time, and keep it in the buffer. This is to align with the disk cluster/block size
<FromGitter>
<stnluu_twitter> For example, even if you're reading just one char at a time, it might just always read a block of up to 1024 chars, and then give you back one char and keep track of where it is internally
<FromGitter>
<stnluu_twitter> double check tho. i didn't look at the underlying implementation, but highly *suspect* that's what it is doing
<FromGitter>
<stnluu_twitter> it's a common practice
<FromGitter>
<stnluu_twitter> on the other note, the delta in perf may be negligible depending on what you're actually doing
Chioque has joined #crystal-lang
Chioque has quit [Quit: WeeChat 2.7]
<FromGitter>
<PenguinOwl> not sure if im doing something wrong, but building crystal release for alpine is taking an indefinite amount of time on `[12/13] [0/1] Codegen (bc+obj)`
ur5us has joined #crystal-lang
sorcus has quit [Ping timeout: 246 seconds]
sorcus has joined #crystal-lang
<mps>
PenguinOwl: it builds in single thread, so yes takes time
<mps>
it takes time on our official builders (Alpine linux) with a lot of RAM and CPUs also
ht_ has joined #crystal-lang
ur5us has quit [Ping timeout: 260 seconds]
alex`` has joined #crystal-lang
asterite has quit [Ping timeout: 250 seconds]
jhass has quit [Ping timeout: 245 seconds]
asterite has joined #crystal-lang
jhass has joined #crystal-lang
<FromGitter>
<btihen> @stnluu_twitter & others interested in Channels and Fibers - I have a very Simple example - with two questions: ⏎ 1) Whats the best wait to wait for all Fibers to finish sending messages (all Channels are empty) and how would that scale with a large chat room? ⏎ 2) Why does switching user who sends the message trigger Fiber switching? ⏎ 3) Would something like this be helpful for the Docs? ⏎ ...
<repo>
i'd say you'd need a status channel. once you break out of the loop you'd send nil to a status channel that was passed to the constructor. Then you just call `USER_COUNT.times { status_channel.receive }`
<repo>
(instead of the sleep
<repo>
(obviously you'd also need to close the channels)
<FromGitter>
<repomaa> you could also do something more elaborate with the status channel, like making it a `Channel(Exception?)` and sending `nil` to it but rescuing exceptions and then sending them to the channel if they occur.
<FromGitter>
<repomaa> this way you can rescue exceptions that happen in a fiber
<FromGitter>
<repomaa> if the amount of users is dynamic, you'd have to set up a separate status channel for each user and then receive on those
Human_G33k has joined #crystal-lang
Human_G33k has quit [Remote host closed the connection]
<FromGitter>
<Blacksmoke16> that approach would work ^
<raz>
the heck, i could've sworn i had tried this before
<raz>
thanks! :)
<repo>
the backslash is key
<repo>
you might've tried without it
<raz>
yes, i probably messed up the escaping with my various "commented out" macro attempts
<repo>
:)
gangstacat has joined #crystal-lang
<repo>
why is Array#clear not Array#clear!
<repo>
it clearly (pun intended) mutates the object
<repo>
same could be asked for #delete and #fill and so on
<FromGitter>
<tenebrousedge> these are good questions
<repo>
i guess there is some consistency though: it seems ! is only used if there is both a mutating and a non-mutating version of the method
<FromGitter>
<tenebrousedge> hmmm
<repo>
if ! is used too strictly it would also mean that []= would have to be []!= or something :P
<FromGitter>
<tenebrousedge> D:
<repo>
maybe that's a bad example, but i see how it could get out of hand
<FromGitter>
<Blacksmoke16> heres a question, working on moving some of my shards into an org
<FromGitter>
<Blacksmoke16> like `myorg/config`, `myorg/logger` etc
<FromGitter>
<Blacksmoke16> however since shards dont get grouped by org, `logger` for example would conflict with other loggers and the stdlibs logger
<FromGitter>
<Blacksmoke16> so my options are like 1) call it `myorg-logger`, or give it a more unique name?
<repo>
i'd call it myorg-logger
<repo>
doesn't mean that the repo would have to be named like that
<FromGitter>
<Blacksmoke16> true true
<repo>
but it would result in a top level namespace Myorg
<repo>
which is nice imo
<FromGitter>
<Blacksmoke16> exactly, thats the plan :)
<repo>
also: if you'r project's name is something like foo_bar. then name the shard myorg-foo_bar
<repo>
so it results in Myorg::FooBar
<repo>
i've done it like this also for rubygems
<repo>
it's a bit work though, because you'll have to rename the shard after creating it with crystal init lib
<repo>
oh wait
<repo>
no you don't
<repo>
just the repo
<FromGitter>
<Blacksmoke16> yea repo would be `myorg/logger`
<repo>
exactly
<repo>
i'm betting the bigger frameworks already do this
<repo>
lucky for example has a lot of sub-shards
<repo>
hm no
<repo>
they have "avram"
<FromGitter>
<Blacksmoke16> yea, they have unique names
<repo>
i'd prefer namespacing it though
<repo>
depending on how tightly they're coupled maybe
<FromGitter>
<Blacksmoke16> now im thinking what to do with the ones that are unique enough, like `event_dispatcher` should it really be like `myorg-event_dispatcher`
<FromGitter>
<Blacksmoke16> i suppose it doesnt really hurt anything
<FromGitter>
<Blacksmoke16> consistency and shows multiple shards come from the same "source"
<repo>
yeah
<FromGitter>
<Blacksmoke16> i could live with that
<FromGitter>
<Blacksmoke16> nice, esp since i can setup `crystal/docs` to build docs for the shards too
<FromGitter>
<tenebrousedge> has anyone written a guide on avoiding "gotchas" in benchmarking Crystal? e.g. cases where calculations will be discarded if the return value is unused, or values will be allocated in an unexpected way
<sorcus>
Blacksmoke16: Hi :-)
<FromGitter>
<Blacksmoke16> hey
<sorcus>
Blacksmoke16: oh, you commented my review. I'm worried and happy. Haha.
<FromGitter>
<Blacksmoke16> yea reminded me i need to handle `HEAD` requests
<FromGitter>
<Blacksmoke16> i.e. if the user defines a `GET` request, define a `HEAD` one behind the scenes that just skips the handler
<sorcus>
Blacksmoke16: too bad i can't help :-( X-)
<FromGitter>
<Blacksmoke16> all good, you can try it when its done and see if you find any bugs 😉
<FromGitter>
<kingsleyh> in a yaml file - can I put crystal code? like: `time: #{Time.utc}`
<FromGitter>
<kingsleyh> do I need interpolation?
<FromGitter>
<Blacksmoke16> can process the file via ECR
<FromGitter>
<Blacksmoke16> interpolation is another option yes
<FromGitter>
<kingsleyh> It's not working :(
<repo>
what did you try?
<FromGitter>
<kingsleyh> I've got a struct with ```timeout: Time::Span``` and then in my yaml file: ``` timeout: #{Time::Span.new(0, 1, 0, 0, 0)}```
<repo>
yeah
<FromGitter>
<kingsleyh> but it says: Error: wrong number of arguments for 'Time::Span.new' (given 2, expected 0, 3, 4, 5)
<FromGitter>
<btihen> @repomaa - thanks. After a bit of playing and experimenting it all works. I assume the call-back on closing is the preferred way - I also realized (after reading your post) I could do: ⏎ ⏎ ``` code paste, see link``` ⏎ ⏎ But I assume that's a bit sloppy. ... [https://gitter.im/crystal-lang/crystal?at=5e0e1b88833c373f4d9a0824]
<FromGitter>
<btihen> @repomaa - would the `delivered?` tell me if there are any messages in the queue before closing? How do I make the channels a bit more `thread-safe`?
Nicolab has joined #crystal-lang
<FromGitter>
<kingsleyh> @Blacksmoke16 thanks :) just what I needed
<FromGitter>
<PenguinOwl> nah its just maintainer prob had crystal installed systemwide on their dev machine
tankf33der has left #crystal-lang [#crystal-lang]
<FromGitter>
<PenguinOwl> path was being passed to make instead of being set as env i guess?
<mps>
no
<mps>
it builds on clean alpine builders, so no crystal installed
Human_G33k has quit [Remote host closed the connection]
Human_G33k has joined #crystal-lang
Nicolab has joined #crystal-lang
ur5us has joined #crystal-lang
<FromGitter>
<btihen> It was mentioned that `Records` aren't mentioned in the Crystal Git book. So I made a pull request with that added: https://github.com/crystal-lang/crystal-book/pull/379 - if you have ideas or feedback -- I'd be glad to hear from you.
<raz>
hm, does crystal spec have something like "let()"? how would i declare a var in before_each so it can be used by the corresponding tests
JuanMiguel has joined #crystal-lang
JuanMiguel has quit [Max SendQ exceeded]
JuanMiguel has joined #crystal-lang
<FromGitter>
<watzon> raz: The built in spec library is pretty bare bones. Personally I really like spectator for when I need to do anything that's not super basic.
JuanMiguel has quit [Quit: This computer has gone to sleep]
JuanMiguel has joined #crystal-lang
JuanMiguel has quit [Quit: This computer has gone to sleep]
JuanMiguel has joined #crystal-lang
JuanMiguel has quit [Ping timeout: 240 seconds]
ht_ has quit [Ping timeout: 248 seconds]
Nicolab has quit [Quit: Leaving.]
alexherbo2 has joined #crystal-lang
<FromGitter>
<wontruefree> we will be hosting a talk about the state of AI in Crystal in 30 min if people are interested in joining
<FromGitter>
<lbarasti> @btihen from your comment on the code snippet above ⏎ ⏎ ``` code paste, see link``` ⏎ ⏎ I suspect you might have misunderstood what `#receive?` does. Just like `receive`, `#receive?` will still block the fiber until a message is available, but it will return `nil` rather than throwing a `Channel::ClosedError` when the channel is closed [https://gitter.im/crysta
<FromGitter>
<lbarasti> Thanks @wontruefree, I was thinking of joining, but it's almost midnight here in the UK, and I'm close to collapsing :D ⏎ Will definitely checkout the recording 👍
<FromGitter>
<wontruefree> sounds good it should be posted soon