Test fails on Windows 10: file copied with sh/copy-file is not the same about spork HOT 13 CLOSED

Not really, when I find time I will go over the current sh.janet in detail. I find the namespace currently a bit unorganized

from spork.

This is probably due to git adding the CR byte on checkout, I'd like to test this, but don't have a working jpm setup on my windows machine.

from spork.

In a different repository I had added this: https://github.com/sogaiu/judge-gen/blob/master/.gitattributes

Does that seem like the kind of thing you had in mind for testing?

from spork.

To be honest, I not really sure. I would have to experiment with it myself. But if that works and doesn't break anything else ok the project, it should be fine I think?
Maybe this should be applied to all static test files then?

from spork.

@tionis

I looked into the situation a bit more.

If test.file doesn't have any line ending characters, the test seems to work fine.

If test.file ends with a newline character, the test fails and it looks like test2.file has both a carriage return and a newline character at the end.

The :w in the file/open of make-new-file may be a factor:

Chaging the mode to :wb made the test pass.

On a side note, it doesn't look like make-new-file's optional parameter mode is used anywhere.

from spork.

I see, I completely forgot about the existence of the binary flag for file/open

from spork.

@tionis

Regarding sh/make-new-file and the mode argument, I filed #98.

Perhaps a straight-forward fix would be to make file/open's second argument mode.

So:

(file/open file-path :w)

would become:

(file/open file-path mode)

If that seems appropriate, perhaps the line in sh/copy-file:

can specify a different mode, say :wb, to address this issue.

How does that seem?

from spork.

Perhaps an additional keyword argument :mode for the file creating functions would be useful instead.

from spork.

By "keyword argument", do you mean this sort of thing? https://janet-lang.org/docs/functions.html#Keyword-style-arguments

from spork.

Yes, exactly, functions using optional arguments with Keywords can be extended later, without breaking previous code and also allow specifying only one select optional argument. While only one would have to be added here, it would be easier to extend later.
I'm curious what others think about this argumentation.

from spork.

For sh/make-new-file, it looked like in ae9db3d, @bakpakin intended for an ordinary optional argument. May be he'd be fine with a keyword argument though.

Regarding:

file creating functions

did you have any other specific ones in mind?

from spork.

bakpakin merged #99 and I verified that the change described in #96 (comment) works on a Windows machine and a Linux machine.

Made PR - #100

from spork.

The test passes for me now.

from spork.