09:09 <dch> servus!

09:10 <dch> I'm porting https://lavinmq.com/ to FreeBSD, and not having success with dependent libraries

09:11 <dch> `E: Locked version 1.1.1 for amq-protocol was not found in git: https://github.com/cloudamqp/amq-protocol.cr.git.`

09:11 <dch> I'd like to avoid fetching the entire git repo for every single dependency (its a huge waste of network b/w)

09:12 <dch> is it possible to tell shards / crystal, hey you can just use the fetched tarball, instead of requiring it to be a full git checkout?

09:13 <dch> for a bit more explanation, the ports build system fetches tarballs for each dependency, which we then cache for future builds, and unpacks them in lavinmq/lib/... where you'd expect them to be

09:14 <dch> but if we're using full git checkouts, this isn't possible

11:24 <FromGitter> <moe:busyloop.net> hmm, you can point shards.yml to a path: https://github.com/crystal-lang/shards/blob/master/docs/shard.yml.adoc ⏎ ⏎ but not sure if that's what you want (it would likely prevent people from building without first manually checking it out to the same path)

11:24 <dch> moe: yep, this is sortof ok for the top-level dependency, but then we have shards that require shards that require shards and it clearly is not an optimal solution

11:25 <dch> I will add this to https://github.com/crystal-lang/shards/issues/523 which seems an appropriate existing issue

11:26 <FromGitter> <moe:busyloop.net> yea it's a known problem (`shards` needs to see tags to identify releases, that's why it does a full clone iirc)

11:26 <FromGitter> <moe:busyloop.net> for most repos it's not an issue as they are tiny - but i can see how it's annoying for large repos

11:28 <FromGitter> <moe:busyloop.net> another thing you could perhaps try is to manually clone only the part/tag you want to `./lib` before running shards 🤔

11:28 <FromGitter> <moe:busyloop.net> but i don't know if that prevents shards from trying to update it with a full pull

11:39 <dch> this was good enough for the moment:

11:39 <dch> - fetch all dependencies as a tarball (no history)

11:40 <dch> run `sed -e 's,github: .*/,path: deps/,' -e '/branch:/d' -e '/version: .>1/d' ${WRKSRC}/shard.yml ${WRKSRC}/deps/*/shard.yml`

11:40 <dch> and then `shards lock` which cleans the mess up

11:41 <dch> not ideal, but enough for today :D

11:48 <FromGitter> <moe:busyloop.net> ah yup, sounds good (if this is for CI then that's prob exactly what i would do as well 😬)

18:02 <FromGitter> <may> Apologies if this has been addressed somewhere already (my google fu failed me today); is there a way to pass variables by reference in Crystal? ⏎ ⏎ For example: ⏎ ⏎ ```code paste, see link``` [https://gitter.im/crystal-lang/crystal?at=63b31c1b410c633d48079c30]

18:09 <FromGitter> <riffraff169> i think youd have to do `user = test`

18:09 <FromGitter> <riffraff169> you could have multiple assign that way too

18:41 <FromGitter> <Blacksmoke16> can use a class for it

18:41 <FromGitter> <Blacksmoke16> otherwise just have the function return the value

19:39 <FromGitter> <may> Yeah, multi assign was where I was headed but I wanted to check first. Thanks folks! ⏎ ⏎ I appreciate knowing the class option as well, thanks!

21:48 <FromGitter> <abduct> hey all, i have a bit of a complex formatting question for an Array(Array(Int32)) variable. the variable consists of all valid [x, y, z] coords for an inverse kinematics solutions generated via (1..210).to_a.each_cartesian((1..90).to_a, (1..50).to_a) do |x, y, z|. since not all coords have a viable solution they are not added to the results. ex: [210, 15, 27], [210, 16, 1] has missing entries. i am wanting to

21:48 <FromGitter> ... format this array so that i have a grid of 210x90 (X x Y) placing the Z value in each cell, but at the same time if there are entries missing, place an X in the cell to symbolize an invalid coord. where do you think i should start?

22:15 <FromGitter> <oprypin:matrix.org> @abduct: couldn't the invalid coord be `nil`?

22:16 <FromGitter> <oprypin:matrix.org> @abduct: can you make a full example of how you want it to look, even with the `X`s that don't actually work?

22:31 <FromGitter> <abduct> that actually sparked a thought. im putting invalid coords as [-1, -1, -1] now. now its just a bunch of iterations through the array to create my grid. i guess i was just caught up on how to identify the invalid results. mental block gone :)