Re: [PATCH v5] Enable xwidgets on macOS

[조성빈 <pcr910303@icloud.com>]

2019. 8. 3. 오후 7:03, Eli Zaretskii <eliz@gnu.org> 작성:

>> From: 조성빈 <pcr910303@icloud.com>
>> Date: Tue, 30 Jul 2019 02:08:40 +0900
>> Cc: alan@idiocy.org,
>> emacs-devel@gnu.org
>> 
>>>>>> +(defun xwidget-webkit-save-as-file (url mime-type file-name)
>>>>>> +  "For XWIDGET webkit, save URL of MIME-TYPE to location specified by user.
>>>>>> +FILE-NAME combined with `xwidget-webkit-download-dir' is the default file name
>>>>>> +of the prompt when reading.  When the file name the user specified is a
>>>>>> +directory, URL is saved at the specified directory as FILE-NAME."
>>>>> 
>>>>> The last sentence is unclear: what will be the directory where the URL
>>>>> will be saved, and what will be the name of the saved file in that
>>>>> directory?
>>>> 
>>>> Well, it was meaning that the URL will be saved at the directory the user specified in the prompt with the name FILE-NAME.
>>>> 
>>>> I’m not sure how I should reword it due to my  poor English skills; can you give me some suggestions? 
>>> 
>>> If you show me an example of using this command, and the results -- in
>>> what directory and under what base name will the URL be saved -- then
>>> I will be able to suggest a better wording.
>> 
>> Sorry for the delay; I was busy for a few days that I couldn't open my mailbox.
>> 
>> Well, while I was trying to give examples, I think I wrote a clearer version of the docstring:
>> "Save URL of MIME-TYPE as file specified by user.
>> When the user specifies a directory, URL is saved with the name FILE-NAME in the
>> specified directory.  The default file name when reading is FILE-NAME expanded
>> with `xwidget-webkit-download-dir'."
>> 
>> May I ask if this is more clear than before?
> 
> I don't think I understand what happens well enough yet.  The command
> asks for FILE-NAME, right?

It’s not an interactive command, more of an helper function.
FILE-NAME isn’t provided by the user; it’s provided by the internal callsite. Maybe embedding the logic instead of making a separate function (which the original patch did) might be better?

>  So when the FILE-NAME provided by the user
> is actually a directory, the URL will be saved in the directory named
> FILE-NAME, but under what basename?

If the user provided directory (which has no relationship with FILE-NAME and is read by the `read-file-name' call inside the function) is a directory, the basename becomes FILE-NAME. The function assumes FILE-NAME provided by the callsite is not a directory.