GNU bug report logs - #16733
messed up unicode chars in package description

Previous Next

Full log

View this message in rfc822 format

From: Stefan <monnier <at> IRO.UMontreal.CA>
To: Juanma Barranquero <lekktu <at> gmail.com>
Cc: Eli Zaretskii <eliz <at> gnu.org>, 16733 <at> debbugs.gnu.org
Subject: bug#16733: messed up unicode chars in package description
Date: Sat, 22 Mar 2014 14:53:31 -0400

>> I think for url-insert-file-contents, the intention is to pretend URLs
>> make up a virtual file-system (that's the point of url-handlers.el).
>> From this point of view http://www.gnu.org/nonexistent is a file that
>> doesn't exist.  So url-insert-file-contents should signal an error just
>> like insert-file-contents would.
> The semantics of a missing file in the filesystem and a missing page
> in a web site are different; filesystems do not usually have a
> "default error file" to return when the search failed; web sites, do,
> and in many cases, that page *is* informative; it can contain helpful
> links, etc. Try accessing

I know.  But we're talking about a file-system, not about the web.
So we're talking about accessing those not in a browser but from any
random piece of code which may do god-knows-what.

>   http://en.wikipedia.org/wiki/Stefan_Monnier
> It returns a 404, but still, the error page created on the fly is
> quite useful (with a simple click you can search all mentions of
> "Stefan Monnier" in the Wikipedia, for example).

That's useful in a browser where the error can be rendered and read by
a user.  But it's inconsistent for example with url-http-file-exists-p.

When your code fetches http://foo.bar/baz.el, it probably won't expect
to get some HTML content instead.

> - Currently, url-insert-file-contents does ignore 404 results and most
> other response codes. I don't think it's a good idea to change its
> long-standing behavior, but if you insist, it's your call.

I think it should be changed, yes.

> - You suggest that url-w-i-n should only call CHECK "in case of a
> problem". I do not like this, because I think CHECK is *the* function
> tasked with telling whether there's a problem or not. How will you
> define what is a problem (what if I really want to get back these 404
> pages?) and how to pass back that information to CHECK?

After url-insert-file-contents is changed, I don't care much about the
rest because package.el will not need to call something else than
url-insert-file-contents so we may not even need url-w-i-n.


        Stefan

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.