11:38 <Groogy> Morning

11:38 <Groogy> o/

12:06 <FromGitter> <aisrael> TIL: If you override a `Foo.new` and don't `yield`, then any code that calls `Foo.new do |x|` silently fails (the block doesn’t get called)

12:10 <FromGitter> <aisrael> OTOH, if you try to override `Foo#foo` but don’t declare a `(&block)` parameter, then the original `Foo#foo()` gets called when you do `instance_of_foo.foo do |x|`

12:14 <FromGitter> <aisrael> (Ok, my understanding doesn’t seem to be 100% spot on. Trying to write some sample code that demonstrates the various behaviour.)

13:42 <FromGitter> <faustinoaq> Hi crystal community, somebody know how can I read updates in a file without printing all file again?

13:44 <FromGitter> <faustinoaq> I'm using `loop { puts File.read("some.log"); sleep 1 } }` but reads all file again

13:45 <FromGitter> <faustinoaq> something like watching a file

13:59 <FromGitter> <cfurrow> @faustinoaq it's not efficient, but you could do `File.read_lines("some.log")` to get an array of lines in the file. On each loop you could also store the number of lines read, and compare it to the previously read number of lines. Then, loop over the array, but only output the line indices that came after the previous number of lines read. ⏎ ⏎ e.g. if you read in a file that had 5 lines, output all 5. on

13:59 <FromGitter> ... the next loop, you read that the same file has 6 lines, so now output only the 6th line. the next loop may have 8 lines, so only read out 6,7,8, or whatever.

14:00 <FromGitter> <cfurrow> I don't know if crystal-lang has a "tail" functionality, but that is what you're really after, I think.

14:09 <lvmbdv> is there a way to "process" a string so "\\n" which is printed as "\n" is turned into an actual newline?

14:10 <lvmbdv> i tried sprintf but didn't work, i think

14:14 <FromGitter> <l3kn> lvmbdv: I'm not sure if there is a method in the stdlib that does this. If not, something like `str.gsub("\\n", "\n")` should work

14:18 <FromGitter> <l3kn> lvmbdv: Maybe even https://carc.in/#/r/3ib3, if you need to handle tabs etc, too

14:19 <RX14> @faustinoaq just call `gets_to_end` in a leep with sleep 1

14:19 <lvmbdv> i need to handle everything C string literals do, just wanted to see if i missed a stdlib function in my search

14:19 <lvmbdv> thank you l3kn

14:20 <RX14> @faustinoaq if you recieve EOF, and then write more bytes to the file, then try to read again you just get the extra bytes.

14:20 <RX14> so you just need to keep trying to read until you get something

14:20 <RX14> unless you want to implement inotify

14:29 <FromGitter> <faustinoaq> > just call `gets_to_end` in a leep with sleep 1 ⏎ ⏎ RX14 Thank you!

14:31 <RX14> i just observed what tail -f did in strace

14:31 <RX14> it's not too hard

14:35 <RX14> @faustinoaq https://github.com/petoem/inotify.cr

14:35 <RX14> you can use this to replace the `sleep 1`

14:36 <RX14> it patches in a nice File.watch(path) do |event|

15:51 <FromGitter> <faustinoaq> RX14 Thank you! I'm trying to add Just in Time output for `amber exec` ⏎ ⏎ ((https://user-images.githubusercontent.com/3067335/35768752-38b06cb6-08cf-11e8-8828-af234c7dd12a.gif)) ⏎ ⏎ https://github.com/amberframework/amber/pull/587#issuecomment-362824321 [https://gitter.im/crystal-lang/crystal?at=5a75da9c98927d5745722f99]

16:51 <crystal-gh> [crystal] ysbaddaden opened pull request #5677: Fix: uninitialized sa_mask value in sigfault handler (master...core-sigfault-uninitialized-mask) https://git.io/vNx3j

17:24 <FromGitter> <drosehn> The `inotify` solution is the best solution. But if that was not available, what you could do is do a `File.stat()` when you reached EOF, and also save the location of `File#pos`. Then each time you wake up from `sleep`, do another `File.stat()` to see if the size of the file has changed. If the file has not changed, then loop back and do another `sleep`. If the size has changed and gotten larger, set

17:24 <FromGitter> ... `File#pos=` and then start reading. If the size has changed and gotten smaller, then you'd probably want to re-read the whole file.

17:27 <FromGitter> <drosehn> I mention this because I have some programs where I don't restart reading from the EOF, but I restart reading from after the last valid entry, where each entry is multiple lines. So it might be that the program does need to re-read some lines, but not the whole file.

17:51 <RX14> @drosehn but that just uses more syscalls then just attempting to read() and getting back 0

17:51 <RX14> the best case: there's more data, you get the data

17:51 <RX14> worse case: there's no data, you get EOF

17:52 <RX14> why do stat when you can just not do stat?

18:33 <FromGitter> <marksiemers> I have crystal built from master locally, if I want to make sure that is used locally for one of my projects, (on macOS) do I just need to switch the symlink here: `lrwxr-xr-x 1 mark admin 43 Feb 2 10:40 /usr/local/bin/crystal -> ../Cellar/crystal-lang/0.24.1_2/bin/crystal` to point to the binary built from master?

18:33 <FromGitter> <marksiemers> Or does something in the shards.yml need to change?

18:39 <oprypin> marksiemers, how about you don't mess with global system files? well on mac it's a lost cause but whatever

18:40 <oprypin> /full/path/to/crystal build stuff

18:41 <wmoxam> will that build against the correct crystal stdlib?

18:41 <wmoxam> might have to muck with some env variables to get that to work correctly :p

18:44 <FromGitter> <marksiemers> Thanks oprypin, I wasn't sure if invoking the binary without changing the system version would use the correct stdlib, etc. (as wmoxam mentioned).

18:47 <oprypin> i dont know how you tie `shards` into this. if it has no way to override the crystal binary, that's pretty bad.

18:54 <RX14> oprypin, PATH="/full/path/to/bin:$PATH" shards lel

19:01 <FromGitter> <drosehn> I thought I had tried "just do another read" and it raised an exception, however that might have been in a different language. All these different programs run together after awhile.

19:02 <FromGitter> <drosehn> Also, I wanted to re-read some of the data. So doing `stat()` is cheaper than doing `pos=` followed by some reads when there is no new data to read.

19:03 <FromGitter> <drosehn> emphasis on "new data".

19:07 <FromGitter> <drosehn> Oh, yeah, it was in crystal, although it was a different program. Not a `tail` program (where I know I have to sleep to wait for data), but in the program where I was trying to duplicate ruby's `popen`. If I did a read after EOF, I got an exception which I then have to `rescue` to get the code to work properly.

19:36 <FromGitter> <marksiemers> FWIW, running the binary directly confirmed what I needed it to: `~/git-repos/marksiemers/crystal/bin/crystal spec spec/amber/router/pipe/error_spec.cr:15`

20:05 <oprypin> ?

21:39 <frojnd> I would like to check if all chars in a string are digits. How would I do that? Something in the line of "123".chars.all? { |c| c =~ [0-9] } ?

21:41 <FromGitter> <bew> you can use `.all? &.digit?` I think

21:43 <frojnd> digit being method of what class?

21:44 <FromGitter> <bew> Char

21:45 <frojnd> digit? for char does not exists

21:45 <frojnd> s/exists/exist

21:46 <FromGitter> <bew> Yes sorry it's number? https://crystal-lang.org/api/0.24.1/Char.html#number%3F-instance-method

21:48 <frojnd> Thank you

23:24 <FromGitter> <marksiemers> oprypin, there was a bug in crystal 0.24.1 and ysbaddaden thought it might be fixed on master, to reproduce, I needed to run a spec in the amber shard. ⏎ That's why I was asking questions earlier about getting a locally built crystal to work with a shard. ⏎ For my case, I didn't need to set the path (per RX14's suggestion) just running `crystal spec` with the locally built binary confirmed that the bug

23:24 <FromGitter> ... was fixed in crystal master.

23:27 <hightower4> Hey, I have a shard in shard.yml which I point to github repo and set 'branch: master'. But, when shard is installed, instead of getting latest master branch, it installs me some version and prints: Installing ...name... (0.9.5 at master)

23:28 <hightower4> How do I make it get the latest thing in master branch?

23:46 <FromGitter> <asterite> branch: master

23:49 <hightower4> Yes, as mentioned I have that

23:49 <hightower4> And it tells me "Installing baked_file_system (0.9.5 at master)"