clojure normally doesn't know which replaceAll method to use there
so it does reflection to see which interfaces filename implements
and it chooses the right method to invoke
with the ^String type annotation it can hardcode that in at compile time
emit the bytecode to invoke that specific implementation
raywillig
so i guess that can save a bunch of time at runtime
michaniskin__
yeah that's what the warning tells you about
raywillig
cool, thx
bsima has quit
NOTICE: [boot] schmir opened pull request #108: throw a descriptive exception when the java compiler cannot be used (master...no-jdk-npe) http://git.io/Fg20
schmir
I'm trying to get boot working on windows...
michaniskin__
schmir: all right!
i know things are rough currently with windows suport
NOTICE: [boot] micha closed pull request #108: throw a descriptive exception when the java compiler cannot be used (master...no-jdk-npe) http://git.io/Fg20
NOTICE: [boot] micha pushed 2 new commits to master: http://git.io/Fgoy
NOTICE: boot/master b18ed9e Ralf Schmitt: throw a descriptive exception when the java compiler cannot be used...
Deraen: I wouldn't do that. too many special cases (on windows).
michaniskin__
schmir: how is windows support for the nio Path class?
schmir
I guess fine. I thought Deraen wanted to do that from scratch!
Deraen
I'm just not very impressed with Javas classes in this case :d
NOTICE: [boot] schmir opened pull request #109: return a File object from relative-to (master...fix-relative-to) http://git.io/FgMb
michaniskin__
also i read that windows can understand paths with / as the separator except on the command line of cmd.exe, which uses the / as the option character, like `ls /l` instead of `ls -l`
Deraen
There was some reason I used URI.relativize instead of Path.relativize. But perhaps it can be used still
michaniskin__
Deraen: i've been using File objects as a way to manipulate paths
which i guess was wrong
schmir
yes, / works fine on windows as a path separator
Deraen
Yeah I think files should be used when you are pointing to some file which might exist. But if your are taking File and using that as path and manipulating path to get relative path, File doesn't point to anything anymore.
NOTICE: [boot] micha closed issue #85: Illegal character error when invoking getting started boot command-line in Windows http://git.io/BaZLQQ
NOTICE: [boot] micha closed pull request #109: return a File object from relative-to (master...fix-relative-to) http://git.io/FgMb
NOTICE: [boot] micha pushed 2 new commits to master: http://git.io/Fgyv
NOTICE: boot/master 056931d Ralf Schmitt: return a File object from relative-to...
yeah so i guess lesson #1 is to use Path objects for path manipulation
?
schmir
Deraen: but that shouldn't be a problem?
michaniskin__: I hope you did test that pull request?
bsima joined the channel
Deraen
Using Paths sounds like proper solution
michaniskin__
schmir: i will test everything before making a release :)
but i think str vs .getPath is ok
i think they return the same thing for a file
i mean "for a File"
Deraen
Looks like relative-to tests are still passing
michaniskin__
schmir: is anything else majorly broken in windows?
and which version of windows are you on?
schmir
michaniskin__: yes
win 7 pro
and server 2008
hard-linking doesn't work, the color codes should also be off by default
michaniskin__
what are you doing instead of hardlinks?
schmir
uberjar doesn't work, since it's trying to copy a file that doesn't exist..
michaniskin__: I turned it off, there's a dynamic var somewhere...
michaniskin__
ah ok, so you're just copying files
that's what happens when the dynavar is off
schmir
windows is such a pain...but I spent the day with boot on linux and only later realized that it doesn't work on windows
michaniskin__
hahaha the ole bait and switch!
well if you tell us what we're doing wrong on windows we can fix it
schmir
anyway, martinklepsch is guilty. he spoke at clojured on boot and that made me try it
michaniskin__
so the Paths thing is #1 lesson learned
schmir
michaniskin__: I think I'll be able to contribute some more pull requests...
michaniskin__
i read that windows did support hard links
it would be a shame if we couldn't use them
schmir
yes, it does. but I got an error message that the file exists. unlinking the target path somehow didn't help...
NOTICE: [boot] schmir opened pull request #110: use deterministic compile order and show progress in aot (master...aot-compile-order) http://git.io/Fg5A
aot compilation doesn't work with the watch task. I'm going to open a pull request in the next days on that one...
NOTICE: [boot] micha closed pull request #110: use deterministic compile order and show progress in aot (master...aot-compile-order) http://git.io/Fg5A
NOTICE: [boot] micha pushed 2 new commits to master: http://git.io/FgFZ
NOTICE: boot/master dff2154 Ralf Schmitt: use deterministic compile order and show progress in aot