(FSPath / 'foo.bar.zot') extension = 'bar.zot'

Hi all,


Currently:

     (FSPath / 'foo.bar.zot') extension = 'bar.zot'

Should it be? I expected it to equal 'zot'.

For reference, in python3:

     os.path.splitext('foo.bar.zot') == ('foo.bar', '.zot')

and in racket

     (equal? (path-get-extension "foo.bar.zot") #".zot")


Tony

Hi all,


Currently:

     (FSPath / 'foo.bar.zot') extension = 'bar.zot'

Should it be? I expected it to equal 'zot'.

For reference, in python3:

     os.path.splitext('foo.bar.zot') == ('foo.bar', '.zot')

and in racket

     (equal? (path-get-extension "foo.bar.zot") #".zot")


Tony

For the 'normal' FileDirectory stuff in the image, the extension would work ok here. I'm not familiar with FSPath - that's the package Colin Putney wrote a while back?


tim
--
tim Rowledge; [hidden email]; http://www.rowledge.org/tim
I came, I saw, I deleted all your files.

On 11/5/20 7:41 PM, Thiede, Christoph wrote:

Good catch! That's clearly a bug IMO. #extension should use #copyAfterLast: rather than #copyAfter:.

On the other, FSReferenceTest >> #testBaseAndExtension explicitly defines a test where it says:

"Note that the extension of a complex extension starts from the first period up until the end"

Worse, it's inconsistent with #withExtension: -

((FSPath / 'foo.bar.zot') withExtension: 'zaz') = (FSPath / 'foo.bar.zaz')

!!!

For the 'normal' FileDirectory stuff in the image, the extension would work ok here. I'm not familiar with FSPath - that's the package Colin Putney wrote a while back?

This is part of package FS-Core that the Squeak config wizard loads as part of the Git support.

Tony

On Thu, Nov 05, 2020 at 06:41:14PM +0000, Thiede, Christoph wrote:
> Good catch! That's clearly a bug IMO. #extension should use #copyAfterLast: rather than #copyAfter:.
>
> On the other, FSReferenceTest >> #testBaseAndExtension explicitly defines a test where it says:
>
> "Note that the extension of a complex extension starts from the first period up until the end"
>

The underlying problem appears to be that someone neglected to write
a test for #extension.

The testBaseAndExtension test appears to be a later addition, and it may
be that the test simply documents an existing bug, so the test itself
may be wrong.

I am attaching a proposed test based on Tony's example.

Dave

On Thu, Nov 05, 2020 at 10:43:02AM -0800, tim Rowledge wrote:
Yes (although the version history seems to show other authors). It gets
pulled in along with the git tools.

It is on GitHub at https://github.com/squeak-smalltalk/squeak-filesystem
which is based on the earlier work at http://www.squeaksource.com/fs

Dave

Am Do., 5. Nov. 2020 um 19:41 Uhr schrieb Thiede, Christoph
<[hidden email]>:
>
> Good catch! That's clearly a bug IMO. #extension should use #copyAfterLast: rather than #copyAfter:.
>

Pharo has it like this as well, so it is also an unnecessary incompatibility:

extension
    "Return the extension of path basename i.e., /foo/gloops.taz
extension is 'taz'"
    ^ self basename copyAfterLast: self extensionDelimiter

>
> On the other, FSReferenceTest >> #testBaseAndExtension explicitly defines a test where it says:
>
> "Note that the extension of a complex extension starts from the first period up until the end"
>

The original author might either have just documented that status quo,
or thought of something like .tar.gz. Again Pharo has changed this to
"Note that the extension of a complex extension starts after the last
extension delimiter".

Am Do., 5. Nov. 2020 um 19:41 Uhr schrieb Thiede, Christoph
<[hidden email]>:
>
> Good catch! That's clearly a bug IMO. #extension should use #copyAfterLast: rather than #copyAfter:.
>

Pharo has it like this as well, so it is also an unnecessary incompatibility:

extension
    "Return the extension of path basename i.e., /foo/gloops.taz
extension is 'taz'"
    ^ self basename copyAfterLast: self extensionDelimiter

>
> On the other, FSReferenceTest >> #testBaseAndExtension explicitly defines a test where it says:
>
> "Note that the extension of a complex extension starts from the first period up until the end"
>

The original author might either have just documented that status quo,
or thought of something like .tar.gz. Again Pharo has changed this to
"Note that the extension of a complex extension starts after the last
extension delimiter".

Am Sa., 7. Nov. 2020 um 13:54 Uhr schrieb Thiede, Christoph
<[hidden email]>:
>
> Are there any objections or shall we integrate this change into Squeak-FileSystem?
>

In my opinion, go ahead, but check and at least post the senders that
need to be changed, please.

On 11/7/20 3:54 PM, Jakob Reschke wrote:
> Am Sa., 7. Nov. 2020 um 13:54 Uhr schrieb Thiede, Christoph
> <[hidden email]>:
>> Are there any objections or shall we integrate this change into Squeak-FileSystem?
>
> In my opinion, go ahead, but check and at least post the senders that
> need to be changed, please.

I don't see any senders (that need changing) at all in my current
trunk-ish image. Are there external packages that need to be looked at
or is it good enough to have trunk be OK for the change?

Tony

Trunk and FileSystem-Git + Squot. Christoph probably has these loaded anyway.

Am Mo., 9. Nov. 2020 um 13:38 Uhr schrieb Tony Garnock-Jones
<[hidden email]>:

On 11/11/20 11:16 AM, Jakob Reschke wrote:
> Trunk and FileSystem-Git + Squot. Christoph probably has these loaded anyway.

The versions of those I have loaded are the ones installed by the
configuration wizard (ie master of Squot and its support libraries) -
and there are no interesting senders of #extension there.

So I think we're probably OK to make the change!

Tony

On 11/11/20 11:16 AM, Jakob Reschke wrote:
> Trunk and FileSystem-Git + Squot. Christoph probably has these loaded anyway.

The versions of those I have loaded are the ones installed by the
configuration wizard (ie master of Squot and its support libraries) -
and there are no interesting senders of #extension there.

So I think we're probably OK to make the change!

Tony