[Date Prev][Date Next][Thread Prev][][Date Index][Thread Index]

Re: language quibbles



On Wed, 17 Aug 2016 22:43:29 -0400, Boruch Baum wrote:
> Finally, if you want any of the content of this email posted to the
> list for wider discussion or for reference, I have no objection. If
> you want me to do so, let me know which sections. If you want to do so
> yourself, that's also fine with me.

I'd like to forward these entire [OFFLIST] mails to the list when
it gets restored.  The maintainer and the chief TSUCHIYA have not
responded yet, though (they academic staff might be in the summer
break).

[...]

> 1] It seems to me the two variables, `w3m-default-save-directory' and
>    `w3m-save-buffer-directory' should be merged into a single variable
>    common to both functions.

Hm, w-d-s-d is a directory used for web parts - images, zip files,
etc.  OTOH, w-s-b-d is for web pages.  I think.  So, I'm not sure
it is a good idea to merge them into a single variable.  Though,
it doesn't matter for me personally how they are, since I don't
use those features so much. :)  See also:

<http://emacs-w3m.namazu.org/ml/msg12385.html>
<http://emacs-w3m.namazu.org/ml/msg12386.html>
<http://emacs-w3m.namazu.org/ml/msg12388.html>

> 2] The default value for those variables should not be "~/.w3m". That
>    folder, in my opinion, is already the victim of bad design by w3m
>    itself in that too much clutter is stashed there instead of in
>    sub-folders, ie. w3mcache*, w3msrc*, w3mtmp*.

>    Possibly better would be either "~/Downloads" or "~/Documents",
>    both of which are standard and generic folders created
>    automatically by many operating systems.

I will not oppose making it the default.  But how do we help
users who want to separate a directory for saving snippets and
a directory for saving web pages?  Or there's no such person? :)

>    To be an even better cross-platform and standards-compliant
>    program, emacs-w3m could take note of the "xfreedesktop" standard
>    on the issue. Xfreedesktop has a specification for defining
>    "user-dirs" (see `man(1) xdg-user-dir'), such as:

>           XDG_DESKTOP_DIR="$HOME/Desktop"
>           XDG_DOCUMENTS_DIR="$HOME/Documents"
>           XDG_DOWNLOAD_DIR="$HOME/Downloads"
>           XDG_MUSIC_DIR="$HOME/Music"
>           XDG_PICTURES_DIR="$HOME/Pictures"
>           XDG_PUBLICSHARE_DIR="$HOME/Public"
>           XDG_TEMPLATES_DIR="$HOME/.Templates"
>           XDG_VIDEOS_DIR="$HOME/Videos"

>    It seems to me to that the current default leads to needless
>    fragmentation od user data among various folders. What I mean is
>    that a user can be expected to use more than one browser, and
>    downloaded html/images should not be scattered based upon the
>    borwser used, but on their nature as downloaded data.

Agreed.  Whatever we do, it would be better to take it up on the
list anyway.

>>> lines 50-51: "You can view the saved page in the \"Next Page\".
>>
>> "Next Page" means a page that will appear by the manu-bar button
>> or the tool-bar button named "Forward to Next Page" (the command
>> `w3m-view-next-page').  So, should it be more definite something
>> like this?
>>
>> You can click the "Forward to Next Page" button on the menu-bar
>> or the tool-bar to view the saved page.

> What's confusing me is:

> 1] The user is saving the currently displayed buffer, so why advance
>    to the next page in the history ring, `w3m-next-page', when she is
>    already looking at the buffer now?

Verify that it's been saved correctly.  The reason I made it be
in the `next page' is that, uhm, I felt it is the right place.

> 2] I may have mentioned before that I use emacs-nox, which in this
>    generation might make my setup unusual. My environment has no
>    modern tool-bar or even old-fashioned menu-bar. As for re-wording
>    the sentence, I would put it in "passive voice" format and refer to
>    the underlying function and maybe even its default keybinding. That
>    might be good enough even for people who use emacs solely in GUI
>    mode because the function name `w3m-next-page' is so similar to the
>    button name. Maybe:

>       "The saved page will be added to the history list, and be
>        viewable using `w3m-next-page'.

I think this is the best, especially for documenting in the doc
string.  Thanks.

>>> Note that saved pages will get shown as what you see in emacs-w3m."
>>>>> I don't understand what this means
>>
>> Some web sites send html data so as to be suitable for an
>> HTTP/1.0 compliant old web browser, that is, emacs-w3m.  So, if
>> viewing the saved data by a modern browser, the appearance will
>> not necessarily be the same as the modern browser that connects
>> to the site directly.

> That brings up an idea... But first: based upon your explanation, I
> would suggest deleting the sentence as unnecessary and obvious,
> because the user is downloading what she is seeing, which is a web
> page presented in emacs-w3m.

> The idea that came to mind is for a command to reload the current page
> using a different user-agent, in order to possibly retreive alternate
> content. That would address what seems to me to be the concern
> underlying the sentence "Note that saved pages will get shown as what
> you see in emacs-w3m.", to wit, that the user might be "missing out"
> by not using 'Internet Explorer' or 'Chromium'. Personally, I wouldn't
> be interested in such a feature, but I'm old, and am the type to be
> using emacs-nox, so I probably don't represent the "average" user
> audience.

Maybe I wanted to put an excuse for the reason why the saved page
that Firefox etc. displays sometimes gets poorer than the one it
usually displays.  I decided to delete the sentence.  Even if it
is important, there is a better place, that is the Info manual.
(Beat me and the emacs-w3m team.  We've been lazy on it for years.)

[...]

> Some comments, such as spelling or grammar corrections, may just be
> clutter to post to the list. Are you the person to whom I should
> directly send that type of feedback, or do you prefer that even that
> be posted publicly?

Yes, please don't hesitate to post such works.  They are always
worthwhile.

> OTOH, what I mentioned above about the
> xfreedesktop standard maybe should be posted to the list because even
> if emacs-w3m doesn't adopt the idea, there is educational value in
> teaching / reminding the existence of the standard.

I see.

> ----------------------------------------------------------------------
> BEGIN: NEW FEEDBACK
> ----------------------------------------------------------------------

> 1] Here are some more proof-reading feedback from my semi-random browsing
> of the codebase:

> 1.1] w3m-lnum.el line 1013: change "poing" to "point"

> 1.2] w3m.el line 5934: change "NO-CHACHE" to "NO-CACHE"

Thanks for the corrections.

w3m-lnum.el was rewritten drastically by Andrey Kotlarski.  So,
I think it's better to take these items up for the discussion in
the list.  (I'm not sure Andrey is still in the list, though.)

> 1.3] w3m-lnum-goto. The description of this function includes the
>      sentence "0 corresponds to location url." Why is this a feature?
>      The user is already displaying the URL in the buffer, and can
>      reload the page using 'R'. Am I misunderstanding its meaning?

> 1.4] w3m-lnum-view-image, w3m-lnum-goto

>      I found usage of these features confusing, and had originally
>      written four long bug reports before I realized how I was
>      expected to use them. My confusion was that the documentation led
>      me to believe that once I type 'I/F', entering an index number
>      would be the *only* way to perform the action. I wrongly presumed
>      that entering 'I/F' a second time would toggle the page state
>      back to normal. This confusion is supported by the response the
>      functions make to pressing a non-numeric key: they remove
>      numbering from links that don't have that non-numeric character.

>      As to documentation, I suggest the following:

> 1.4.1] w3m-lnum-goto

>        "Display lnum-style numbering for links, images and forms, and
>        then prompt the user in the mini-buffer for which to open or
>        display.

>        The user response may be an lnum number or may be any unique
>        substring of the text label of the link, image or form. In all
>        cases, the lnum numbering will dynamically update in the buffer
>        as one types one's response in the minibuffer."

> 1.4.2] w3m-lnum-view-image

>        "Display the image under point in the external viewer.

>        If no image at point, display lnum-style numbering for images
>        and then prompt the user in the mini-buffer for which to view.

>        The user response may be an lnum number or may be any unique
>        substring of the text label of the image. In either case, the
>        lnum numbering will dynamically update in the buffer as one
>        types one's response in the minibuffer.

>        `w3m-content-type-alist' defines the external viewer and
>        parameters to be used for any mimetype. `w3m-image-viewer'
>        defines the default image-viewer."

> 2] Feature request: w3m.el: line 1139: Please define variable
>    `w3m-image-viewer' using "defcustom".

Ok.

> 3] Suggestion: Variable `w3m-content-type-alist' could use external
>    program "xdg-open". As mentioned earlier in this already too-long
>    e-mail, there exists a cross-platform xfreedesktop standard to try
>    to assist developers of individual programs to offer users
>    consistency *between* programs. One part of the standard offers a
>    framework and programs "xdg-mime" and "xdg-open" to define and
>    respect a user's preferences across all that user's programs. Thus,
>    if a user changes a preference from within "nautilus" file manager,
>    or from within any other program, that change would respected by
>    all other programs that are compliant with the xfreedesktop
>    standard.

AFAIU, xdg-open is problematic, especially in the case using it
with Emacs (I'm a Fedora Linux user as well).  It passes operands
to a real viewer and returns immediately, so there is no way for
Emacs to know when the external process really finishes, and when
temp files can be deleted, etc.  But I can think it is good to use
such a standard program, too.

>    An example of a command-line program that uses this standard to
>    great effect is "Midnight Commander".

>    You know what? It just occurred to me that emacs, GNU, and the FSF
>    may have issued policy or guidance about xfreedesktop. I've already
>    spent way too long on this e-mail to be motivated to check today.

I'll install (some of) your contribution tomorrow.
Thanks again.

Regards,