Fork me on GitHub
#boot
<
2015-11-01
>
taylor.sando00:11:38

I believe you control the output directory with :compiler-options {:asset-path "/path/you/want"}

martinklepsch00:11:48

@domkm: things being written to out in your project dir can only be done by output options. I'd check there. Gotta sleep now. Talk later!

domkm00:11:53

@martinklepsch: Output options in that pipeline?

sjm10:11:42

sigh, just to let you know that my supposed deterministic system is being thwarted by logic in boot. I'm getting this error "Please download latest Boot binary: "

martinklepsch10:11:45

@domkm: there were no options in your snippet but I guess somewhere in the stuff you trimmed maybe?

martinklepsch11:11:13

@sjm: that should only happen when you updated boot and didn't pin a version in your project.

sjm11:11:33

Okay, I turned on my machine, ran nix-shell (after running it yesterday) then ran boot dev the exception threw. I tried changing my nix scripts to point to the new boot_bin but I got the same error. How exactly do I pin a boot version to my project? On the train atm sorry for the vague response.

martinklepsch11:11:45

@sjm: boot -V prints a thing that you can put into boot.properties

martinklepsch11:11:01

If there's a boot version given there it uses that one.

martinklepsch11:11:20

I think new boot-bin is not yet officially released

martinklepsch11:11:35

Try 2.4.2. Should work with any the latest boot.sh from the boot project or boot-bin

alandipert13:11:35

anyone headed to oredev this week? i'll be there tuesday, would love to meet up with boot users

sjm13:11:37

@martinklepsch: ah thanks, I updated the nix scripts to point to the new boot_bin properly. Boot works now.

kurtlazarus15:11:45

hey guys, just started using boot - trying to use devcards with it: https://github.com/bhauman/devcards#usage-without-figwheel

kurtlazarus15:11:56

where should the build-options go?

micha15:11:56

you probably don't need them

micha15:11:33

that's some stuff to inject things into cljsbuild

micha15:11:38

or something

kurtlazarus15:11:43

eh, i’m halfway there - i got the devcards title to show up, but none of my cards are rendering

onetom15:11:51

A boot upgrade can fail in a few ways... 😕 I've Ctrl-C'd it because I thought if I ask it to upgrade, then it would download 2.4.x, but instead any consecutive runs fail with the error shown:

> boot dev
Downloading                                                                                                   

> boot -u
Exception in thread "main" java.lang.ClassNotFoundException: boot.App
	at java.net.URLClassLoader.findClass(URLClassLoader.java:381)
...

micha15:11:44

onetom: can you re-download boot.sh?

micha15:11:53

that was fixed

onetom15:11:39

i was installing boot with brew install boot-clj. i tried to uninstall and install it again, but that doesnt help

onetom15:11:03

i guess if i would wait for the 2.4.2 formula released, that would fix the issue

micha15:11:07

yeah i don't thnk brew is up to date

onetom15:11:59

ok, nvm, i will use a manually installed one for now

micha15:11:50

we have a fake "latest" branch so you can download from the same url to get the latest thing

micha15:11:57

without needing to know the latest version

micha15:11:10

there should never be a need to use an older version

micha15:11:52

there are also real tags and releases in case you want to package it and need a versioned url

domkm19:11:54

@martinklepsch: That's the complete task. I didn't think anything I previously left out would change cljs compilation.

domkm19:11:34

@martinklepsch: btw, the "out/" dir gets created when I connect to the nrepl server and type (start-repl) to turn it into a cljs repl

juhoteperi19:11:18

Probably a problem with boot-cljs-repl not setting all all the options

martinklepsch20:11:26

@micha: Good sunday afternoon!

martinklepsch21:11:43

@micha: you mentioned you had something you wanted to push?

martinklepsch21:11:57

like something related to the boot builds boot stuff

micha21:11:33

i ran into issues deploying the test

micha21:11:36

not sure what it was

micha21:11:41

something with the pom was not right

micha21:11:53

crashed when boot tried to read the pom to deploy

martinklepsch21:11:45

you mean like deploying to clojars?

micha21:11:00

it installed to local maven just fine, which is the weird thing

micha21:11:32

:thumbsup: for homebrew

micha21:11:31

@martinklepsch: i added a thing today to let you do boot show ... on any pod

micha21:11:49

would have been helpful for debugging the boot-cljs issues

martinklepsch21:11:19

that sounds great. how do you point to a pod? or does it just print it for all?

micha21:11:31

the default behavior hasn't changed

micha21:11:44

but the -P, --pods option has been added

micha21:11:55

where you give it a regex that will be matched against the pod names

micha21:11:07

pods are named after the namespace where make-pod was called

micha21:11:12

so you can do like

micha21:11:28

boot cljs show -dP boot-cljs

micha21:11:35

and that will show you the deps tree of the cljs pod

micha21:11:53

boot hoplon cljs reload show -dP .

micha21:11:07

to see dep tree of any pod (because . matches anything)

micha21:11:22

works for show -u, show -U, show -p and show -e

martinklepsch21:11:36

how does it work with multiple pods from the same ns?

micha21:11:42

it will show them all

micha21:11:06

if you want you can name a pod yourself by calling .setName("asdf") on it

micha21:11:39

like if you are making lots of pods, or if you are making a library that will handle pod creation on behalf of other libraries

micha21:11:42

i was talking to alan today about using lucene for the sift task

micha21:11:54

and for things that need to find files in the fileset

micha21:11:08

also more pervasive caching

martinklepsch21:11:38

caching for what?

micha21:11:48

caching of blobs

micha21:11:56

to reduce copying

martinklepsch21:11:09

when are they dumped from the cache?

micha21:11:25

when the cache runs out of space probably

martinklepsch21:11:42

how can you make it more pervasive than that?

micha21:11:48

it would require a mark and sweep kind of situation

micha21:11:16

that would eliminate most of the work done with files in the fileset

martinklepsch21:11:25

I don’t think I understand. Isn’t most stuff linking anyways? Things are only copied to the blob store once?

micha21:11:31

yes but that could take a long time if there are a lot of things

martinklepsch21:11:55

but only ever once when starting the pipeline

martinklepsch21:11:08

are you talking of caching for this startup phase?

martinklepsch21:11:24

ok, makes perfect sense then

martinklepsch21:11:10

@micha: re that deploy issue: you made sure you’re not building the version of boot you’re using?

martinklepsch21:11:33

How did you test the deploy? pushed to some other groupId?

micha21:11:05

i was pushing a snapshot

micha21:11:25

i want to get these latest changes deployed before trying again

martinklepsch21:11:35

@micha: doesn’t handle things on a per dir basis but I guess that’s an acceptable tradeoffs as dirs don’t change usually..

micha21:11:46

how do you mean?

micha21:11:51

the fileset dirs never change

micha21:11:09

i mean you never add or remove dirs

martinklepsch21:11:41

In Boot, but in theory it’s not impossible right?

micha21:11:00

yeah i suppose it's possible

micha21:11:14

you could accomodate that easily though

micha21:11:26

since adding a dir just means updating the state

micha21:11:39

so we'd have an add-dir protocol method

micha21:11:47

that would update the atom

martinklepsch21:11:50

but then the set would be different and everything would need to be linked again

martinklepsch21:11:05

ah you mean add the dir to the key

micha21:11:11

no, it would just add the dir to the key yeah

micha21:11:43

i don't see any use case for that though

martinklepsch21:11:56

yeah I suppose that would work. Felt more general to just key by dir but I like the size if your diff more simple_smile

martinklepsch21:11:03

Pull in the tests though simple_smile

micha21:11:49

i had some fileset tests once upon a time lol

martinklepsch22:11:52

@micha: wondering if you have thoughts on merge commits vs. rebased fast forward merges? I usually don’t see much utility in merge commits, especially if it’s just single commits.

micha22:11:52

i don't have a strong opinion either way

micha22:11:20

some people mind seeing the merges

micha22:11:06

i do enjoy squashed commits when appropriate

martinklepsch22:11:53

Alright. Good night! simple_smile

alandipert23:11:12

does anyone have thoughts about what a presentation of boot to non-clj/cljs users should include/not include?

alandipert23:11:34

i am presenting this week at a general dev con and the talk introduces the concept of programming with values, then introduces fileset

alandipert23:11:29

(in the process of prep i realized how hard a presentation of boot to people who don't know clj will be, am a little scared)