Browser-based file downloads (was: Re: [vox-tech] browser and mimetypes (I think))

[Karsten M. Self]

on Sun, May 08, 2005 at 10:56:54AM -0400, Peter Jay Salzman (p at dirac.org) wrote:
> There's an URL to a pdf download of an article published by Optics Letters.
> I don't think anyone can access this because I'm using a Princeton
> University proxy (and Princeton has a subscription to this journal), but the
> URL is:
> 
>    http://ol.osa.org/ViewMedia.cfm?id=68550&seq=0
> 
> It doesn't look like a pdf.  I tried and tried to get the pdf to download
> and display on Firefox on Linux.  I think perhaps it downloads OK but
> Firefox doesn't know what to do with it, since it's a "cfm" file, whatever
> that is, not a pdf file.

FWIW, my browser of least suckitude currently is Galeon.  It uses the
".desktop" file method for determining what to do with various online
media.  This has been broken for months (and I've got no clue and not
much inclination to fix it).  It's documented, in very Joyceian fashion,
at http://www.freedesktop.org/.

Moreover, Galeon (1.3.x) has changed from the 1.2.x mode of "what do you
want to open this with, here's a list, select from it, or add your own
handler, and tell me if you want it to be a default" to "Um.  Somewhere
buried in the bowels of GNOME We Who Know Better Have Determined That
This Is What Thou Shalt Use".  Theoretically this is modifiable ... by
firing up that not-quite-a-file-manager / not-quit-a-window-manager
abortion, Nautilus.  Which inevitably screws with my desktop, xmodmap
settings, and other issues.  I can't be arsed.

So, I fire up a terminal (<shift>-<alt>-t in my WindowMaker
keybindings), 'cd /tmp; wget -O foo.pdf "<url>"'.  Note that quoting
URLs is required for anything containing '&' arguments.  Note too that
this breaks for crap that uses redirections _after_ the URL (Sun's Java
docs apparently do this -- if you want to see me red in the face, ask
about the pain of installing Java on Debian...).

Still, it beats d/l'ing the wrong file, not having a reader launch,
and/or trying to figure out where the hell the damned thing saved to.

Firefox, FWIW, apparently launches nautilus by default, from its
download manager dealiebopper.  Same fails to accurately identify the
directory to which files are downloaded, and fails to provide user with
choice of where to d/l a file, at the time of d/l (it's configurable in
advance through preferences, somewhere).

Usability disasters all.


The other issues you may have to deal with:

  - Browsers apparently use the MIME type information concerning a file
    to figure out how to display it.  See above for the .desktop
    clusterfsck WRT GNOME.  Used to be something you'd manage w/in the
    browser.  Oh, except when they don't.  And MSIE apparently has
    hardcoded filetypes based on extensions, so for the 95^W90^W85^W 80%
    of the Web using the Worlds Worst Webbrowser, behavior is completely
    different.

  - As noted above, some sites (mostly corprate fsckups dependent on
    Java or Javascript) don't link directly to downloads, but to an
    intermediary page.  Sometimes it's trivially possible to determine
    the proper URL, sometimes not.  Worse:  d/l managers from Firefox
    and Galeon (GNOME) fail to provide the specific URL currently being
    fetched in a form that allows ready cut'n'paste to a terminal.  Or
    at all.


Peace.

-- 
Karsten M. Self <kmself at ix.netcom.com>        http://kmself.home.netcom.com/
 What Part of "Gestalt" don't you understand?
    I really do not / want to take this step, but he / is forcing my hand.
    - Haiku forced hand.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://ns1.livepenguin.com/pipermail/vox-tech/attachments/20050508/647dc890/attachment.bin