A module to expose all commands in PATH

[chrisdone]

I've been experimenting with this approach, using template-haskell.

It allows you to just write git ["status"] or date [] or rm ["file"]. For me this changes scripts from "this is a bit tedious" to "this feels more bash-convenient".

Would you be interested in adding such a thing to shelly? It takes some seconds to compile the module (maybe 10) that uses generateBinaries, and makes a rather large haddock output, so you might want to make it a flag or something.

[gregwebs]

Interesting! Shelley already has code for searching the PATH. Why does it need to be flagged? Shouldn't it just take time to compile when the user uses it?

[chrisdone]

Interesting! Shelley already has code for searching the PATH

Sweet. I just cobbled mine together, I expect you'll have something more honed and battle-tested. Can you point me to that code?

Why does it need to be flagged? Shouldn't it just take time to compile when the user uses it?

Oh, was just a thought. It's indeed just compile-time, just adds a bit of time in compilation. If you're cool with that by default, so am I.

[chrisdone]

Hold that. I'm cloning, I'll have a look around and make a PR for this.

[gregwebs]

https://github.com/yesodweb/Shelly.hs/blob/master/src/Shelly.hs#L573

[gregwebs]

But I don't get the by default bit: doesn't a user choose whether to use this?

[chrisdone]

But I don't get the by default bit: doesn't a user choose whether to use this?

I just meant that it should build this large module. Nevermind, I was making a big deal out of the build time, it's not that bad.

Anyway, I suppose someone would write:

import Shelly.Bin

And that module brings a bunch of git = cmd "git", etc. top-level bindings. What do you think?

[chrisdone]

https://github.com/yesodweb/Shelly.hs/blob/master/src/Shelly.hs#L573

Right, so can I move the code at here to the top-level as an IO function that can be used from within TH?

[chrisdone]

I.e. to get a function like this.

[gregwebs]

sounds good! If it just works on unix for now that is ok (there is the .exe bit at least)

[chrisdone]

Cool. Alright, I'll make a PR later this evening!

[gregwebs]

Looking at this more I feel that there are some downsides to the approach as is

1) additional build time and binary size 2) scope pollution, which requires a remap for things I don't even care about and makes the script non-portable 3) the PATH can change from build time to run time

It seems like the issues can be solved by stating which binaries you desire: a white-list api:

generateCmds ["git", "less"]

For point 3, the script no longer has to check for the existence of a binary, although it could warn about the binary not existing.

I can see how you wouldn't want to take this approach in a shell like Hell though.