#hoplon

anybody checked this out: https://github.com/amatus/gnunet-web ?

pandeiro: yes, i still maintain secretary. there's pr for the next major version but it hasn't been merged yet because i haven't had the chance to update the README for it.

martinklepsch: ping

pong

for cljsjs libs that depend on other cljsjs libs, we need to both include the dependency and :requires, is that it?

eg react-router should have cljsjs.react 0.12.x in deps, and :require cljsjs.react as well?

(i just realized i haven't been including the dependency part, just the :requires)

yup that sounds right

ok i gotta fix some packages :)

so this means that package A is pegged to a version of dep package B

(unless it's overriden in the application's top-level dependencies)

yes I think so

yep ok makes sense

martinklepsch: thanks, that is all :)

pandeiro: cool :)

Does Boot create "target" if it doesn't exist?

pandeiro: yes

pandeiro: but only at the end of the cycle

pandeiro: (I assume you're asking because of these boot-http issues)

martinklepsch: yes exactly

martinklepsch: so is there any way boot-http could tap into that beforehand?

i guess not?

pandeiro: don't think so

yeah, hmm

alandipert: suggested just creating the directory that's passed to :dir — seems like the best thing to do to me

could boot not do that at the beginning instead of the end?

martinklepsch: `boot serve -d tagret`

tagret is created silently and served empty

nobody notices until you're demoing

:)

that is my concern with that

well.. you can't account for typos really

you could say 'tagret doesn't exist, quitting'

or maybe special-case target

pandeiro: maybe introduce an option to create the dir if it does not exist

is target the official target btw?

pandeiro: it's the default

being the default maybe we can special case that one, otherwise exit 1 w/ an error

alandipert: ^-- sound sane?

or maybe stay away from special cases and make the directory but print a warning

yhe target can't be created eagerly though

it needs to be created lazily at the end

and only if there are files to put in it

michaniskin: ah ok, i'm sure you've thought about it

michaniskin: any take on the options above?

like should `boot -h` create a target dir?

michaniskin: ha

why not right?

tasks shouldn't be looking in there

the answer to the boot-http problem is to use temp dir instead of target

michaniskin: so don't actually serve the directory the user specifies, copy all of its contents to a tmpdir and serve that?

make a temp dir and in your task copy all output files to it

michaniskin: and for adhoc `boot -d pandeiro/boot-http serve -d . wait` too?

(there is no concept of an output file there, right?)

no build.boot, etc

right

but even then

maybe better would be

boot -r target -t /dev/null serve

boot needs a -T / --no-target option to set --target /dev/null

michaniskin: but 'better' in what sense?

for the end user, i think `boot serve -d target` is better, no?

but its error prone

yeah?

because you're defeating the pipeline model

yeah i see that

target is in flux

when using boot in a pipeline

so when you start 2boot processes like for a repl client

they fight over the target

things like that

yeah that makes sense

granted i haven't heard people mention that wrt http/serve 'in the wild'

have you hit that?

there are some issues in the discourse site about it

k i will have a look and study this further

maybe we should use file locking on the target dir