Clojurians Log v2

@bhagany I think you might be really onto something, I remember having this: https://github.com/pandeiro/boot-http/issues/70

It looks like I misinterpreted the problem and your explanation is probably what's happening

@richiardiandrea indeed, trying to use boot-http is also how I discovered it 🙂

@bhagany so I will need to fix it too very soon cause now I remember why I ran into that perun "misunderstanding" that you fixed. I tried to move the files back in resources and puff...now boot-http is broken because of the above 🤦 Thanks for digging btw, I did not have time at the time.

I’m hoping to open a PR tonight

@richiardiandrea interesting discovery: (sift :move {#"foo\.txt" "bar.txt"} :to-asset #{#"bar\.txt"}) sticks

oh I never ever use that feature

sift?

no I mean :to-asset

I never have either

but it is a good point, I always assume that sift does the right thing and uses the output roles...

maybe not?

I don't think it does

I was wondering why when I do something like (-> fileset (boot/add-resource tmp) boot/commit!) it'll stick, but sift or mv won't

and now I think it's because boot needs to know if it's a resource or asset or whatever

@bhagany from the code both :move and :add-* use boot.core/get-add-dir

digging more

I'll be very glad to have other eyes on it 🙂

I don't see where sift :move does get-add-dir, though

sorry actually no, sift-action :move uses another code branch

cool, we're on the same page

yeah that's tricky because sift-mv is used there on top in those defmethod

very interesting that :to-* wrap ... exactly

maybe :move should use the same code, sift-mv, with a default role ?

I've been wondering if there's a way to easily determine the role of the src file...

I would default it to one that outputs

I feel like it's hard to pick a default for it

@bhagany uploaded a file: https://clojurians.slack.com/files/U051706GF/F8P81HUSX/image.png

well 80% of the times you want to transform from an input role to an output role, or from a resource (which is output atm) to output

that ?? gets me every time

yeaaaaaaaaaaaaa

so I would default to asset

hrm, I was leaning toward resource

at least in the case of Perun, we often want to use files as input as well

after processing them a bit

right

you are right boot is compositional/transformational...so everything that is more than just one task will need :resource-like role

notably only the last task has input->output

yeah, that's is how I understand it as well

not that I'd be very surprised to find that I don't actually understand it 🙂

I am trying to use sift-mv in place of the current :move implementation

I'm not sure we can do that

sift-mv modifies :dir, and sift-action :move modifies :path

my first impulse is to modify sift-action :move so that it also modifies :dir

yes somehow :dir needs to be added in there

https://gist.github.com/bhagany/2bfd183d768487cdd2719bddd58a2af6

^ that works for me

... nope, taking that back

I was about to try, but ok, no I am taking back the attempt 😄

sooo @bhagany also sift reduces over actions so basically if I understand correctly first :path is modified then :dir

(maybe you've noticed already :D)

https://github.com/boot-clj/boot/blob/master/boot/core/src/boot/task/built_in.clj#L631

do you mean for something like (sift :move {#"foo\.txt" "bar.txt"} :to-asset #{#"bar\.txt"})?

yes, so :move is done before :to-asset

it seems

yes, but I think that might just be luck

I don't think the order of reduction is necessarily defined there

Also, I modified the gist to something that actually works 🙂

if you see the line above there is a reduce-kv

oh cool

another fix would be to compose the two, if this reduction theory is right 😄

possible, but I didn't really think that we would need both reductions over the same fileset

well yes actually don't want to slow you down if you have another working solution

well, fwiw, I am not terribly confident that this is the right thing to do. I feel like there might be some reason for the current behavior that I'm not seeing

and also, I can now work around it

hrm, probably worth an issue, possibly pinging Misha for some insight

yeah, I'll open an issue, and perhaps a maintainer or two will take the trouble to read this conversation 🙂

😄

maybe I'll skip right to the PR and see how it goes

Would be awesome, also to see if the current tests are passing (even if not very many)

yup, tests pass

on os x at least

Anyone comfortable with Java could have a good impact here: https://github.com/boot-clj/boot-bin/issues/6