@alandipert: thanks for your answer to my second question. I should have made my first one clearer: Let's say I have a Boot pipeline of three tasks: A, B, C. Most of the time all of them work, but at one point of development B throws an exception. Will C still get called?

@myguidingstar: no

then if the REPL task is after another that throws exception, then will (load-file ...) work?

hmm, maybe I mess things up, REPL should starts it server instance instead of being inside a pipeline life cycle

oh, usually what i do is boot repl, then at the repl to run the build i do (boot (A) (B) (C))

since the repl isn't really part of my build, it's just a way for me to direct the build

so things like boot dev is not recommended?

i think in most cases it's fine

anyway it's good to develop different workflows, i definitely use boot differently depending on the kind of application

I see

hmm, if a task before boot-cljs-repl throws exception, then boot-cljs-repl should hang, no?

i'm not sure exactly how cljs repl works, but the builtin repl starts up when the task is constructed, and is a no-op when the fileset gets passed

the repl has always been a weird thing to think about in a pipeline, it doesn't really fit the idea

since its purpose is to facilitate interaction NOT via the fileset

so it's not clear to me how the repl should behave in that circumstance

yeah, it's the repl in pipeline context that confused me, not Boot pipeline itself

so build.boot with REPL tasks should be like this:

(start-repl-server) (start-cljs-repl-server) (boot (notify-repl-server) (notify-cljs-repl-server) ...)

any idea?

hm, i just had a thought

kind of inspired by the way you have symmetric tasks

so the REPL isn't a good fit for boot builds because it's a separate pipeline

but what if we decomposed the REPL pipeline into parts and made each a boot task

so instead of an nrepl client, we have a task that listens for expressions and runs the build pipeline with that form added to the fileset

and then later in the pipeline, a separate task that evaluates it in the right context

i guess really the "repl" could be a program that writes code to your source dirs

and then all the normal boot/watch stuff is responsible for eval

exactly

that's what I feel missing

eval in the pipeline context!

i have to go but if you have any other thoughts on this i'm very interested to hear them

yes

this is how i do cljs personally, i haven't used a cljs repl since 2012

live reload ftw

this would line up repl experiments with the actual app environment also, no context switch

source of many problems and misunderstandings

right

it's like Lisp & Boot have it all but we mess 'em up

well i'll see you later, thanks for stimulating discussion

haha yes

I am rewriting a lein project in boot, and I am having some trouble moving over some code that ran my datomic migrations. Before, I simply passed a sorted list of each migration filename to the conformity read-resource to get back the tx data, that was then passed along to ensure-conforms which transaced the data. Each migration is in resources/migrations directory. It seems that ( file-name) is returning nil despite the file being valid and in the resource dir.

https://github.com/rkneufeld/conformity/blob/master/src/io/rkn/conformity.clj#L22

Here are my src and resource paths:

:source-paths #{"src"} :resource-paths #{"resources"}

@kenbier: if I remember correctly io/resource is used for fetching within jars, what I usually do for targeting both repl and jar is:

(or (io/resource "version.properties")
                   (io/file "version.properties"))

then slurp or io/reader

hmm, let me try that

well that does seem to get me a step futher

passing in just the filename in the lein repl worked but in boot it doesn’t. so im going to try passing in the path relative to resources/ and see if that makes a difference

now io/reader is complaining the file cannot be found

if you have it in the resources you should see it on the classpath under /

if you do boot show -f you can also see your fileset

and see where it is/will be placed in the uberjar

(it can be big to display though, so a grep is good there)

migrations ├── 1453359339862_navi-app-schema.edn └── 1453359390119_navi-app-data0.edn

looks like its under /migrations ?

yeah

so in lein my resource path looked like this: :resource-paths ["resources" "resources/migrations"]

yes that should be the same as in boot

that concept actually comes from maven

well it complained about the files overlapping, so my resource-path is just [“resources”]

perhaps lein was putting the files on / and not /migrations?

if you put stuff in resource-paths in both boot and lein

you will see them in /

so now in boot you have only resources, therefore the content is mapped to /

this is why you see them under migrations

that makes sense, thanks

np 😉

btw if you are using ragtime I suggest you to have a look at boot-ragtime

i considered it but am not. have you used it with datomic before?

ah not with datomic, only postgres

yeah if i were using postgres i think id surely be using it