fix: add --output to wat2wasm, wasm2wat and opt cmd

[vasucp1207]

Emit wasm files in same dir as wat files

[zapashcanon]

Also, I'm not sure we really want this behaviour, I checked and the wat2wasm command from wabt actually outputs the file in the current folder. @filipeom, @krtab WDYT?

[vasucp1207]

Also, I'm not sure we really want this behaviour, I checked and the wat2wasm command from wabt actually outputs the file in the current folder. @filipeom, @krtab WDYT?

Yes, you are right, because the wasm2wat emit the files in the same folder that's why I did this.

[zapashcanon]

Also, I'm not sure we really want this behaviour, I checked and the wat2wasm command from wabt actually outputs the file in the current folder. @filipeom, @krtab WDYT?

Yes, you are right, because the wasm2wat emit the files in the same folder that's why I did this.

Aw OK. Maybe we could have a flag then, something like --emit-file-in-current-directory for the same behavior as wat2wasm and keep the new one as default; or --emit-file-in-same-directory to get the new one and keep the wat2wasm one as default ?

[filipeom]

Also, I'm not sure we really want this behaviour, I checked and the wat2wasm command from wabt actually outputs the file in the current folder. @filipeom, @krtab WDYT?

I understand why this PR is needed, but I'm not entirely in favor of it. I believe both --emit-file-in-current-directory and --emit-file-in-same-directory are adequate solutions.

However, I must confess that I prefer the wat2wasm approach: passing in one file and exporting it to another using -o. When I need to process multiple files, I can simply iterate through them in a shell loop. I think accepting a list of files as input is less flexible, as there might be use cases where someone wants each file to be exported to a different destination.

I'll leave the final decision to someone with a stronger opinion.

[vasucp1207]

However, I must confess that I prefer the wat2wasm approach: passing in one file and exporting it to another using -o. When I need to process multiple files, I can simply iterate through them in a shell loop. I think accepting a list of files as input is less flexible, as there might be use cases where someone wants each file to be exported to a different destination.

I'll leave the final decision to someone with a stronger opinion.

I agree that accepting a single file as an argument is a good idea for specifying the target path, let wat2wasm be the default approach and I suppose there is no need for these type of flags --emit-file-in-same-directory.

[zapashcanon]

@vasucp1207, do you want to make the changes to accepting a single file, to use the current directory and to accept the -o option? :)

[vasucp1207]

@vasucp1207, do you want to make the changes to accepting a single file, to use the current directory and to accept the -o option? :)

Yup, wabt also accepts a single file.

[zapashcanon]

Yes sure, so I propose the following changes:

• make owi wat2wasm and owi wasm2wat accept only one file;
• add a -o FILE option to both (we can keep --emit-file too for owi wat2wasm);
• make the default output directory be the current directory (when -o FILE is not present).

[vasucp1207]

• make the default output directory be the current directory (when -o FILE or --emit-file is not present).

Not include the --emit-file option to wat2wasm because it's by default emits the file anyway.

[zapashcanon]

@vasucp1207, I believe we also have to do the same for owi opt along the way

[vasucp1207]

@vasucp1207, I believe we also have to do the same for owi opt along the way

Sure, but I didn't add --emit-file to owi opt because of name conflicts if the source file is in the current dir

[zapashcanon]

This should be updated now :)

[zapashcanon]

We are getting there! Don't forget to run and promote the tests once the last issues have been fixed so that I can merge directly. :)

Also, could you add a test for each command (they each have a directory in test/) ? It could be as simple as:

$ owi {opt,wat2wasm,wasm2wat} foo --output=bar
$ owi fmt bar

[zapashcanon]

Thanks for your patience! :)