Re: $BROWSER

[Pavel Roskin <proski gnu org>]

Hello, Eric and Mathieu!

> I was trying to point out the fact that when users set explicitly
> their browser to konqueror (for instance with gnomecc) $BROWSER remain
> unset, which in anomalous IMHO.

Yes, I fully agree with that.  If there is a default browser for the
desktop environment and all programs native to that desktop environment
already obey that setting, and that desktop environment is currently
active, then the BROWSER environment variable should be set to communicate
the browser preference to other (i.e. non-native) software.

I believe that the BROWSER proposal should be added to the standards at
freedesktop.org.

> It seems overcomplicated to me to define BROWSER as
>         BROWSER="netscape -raise -remote \"openURL(%s,new-window)\":lynx"
> while a simple
>         BROWSER="mozilla" or BROWSER="galeon --new-window"
> can work, as most recent browser understand a command like
>         $BROWSER $url/$path

I think I have seen this criticism already.  Indeed, the requirement to
support colons and %s add complexity to the implementation.  For example,
GNU Midnight Commander uses a configuration file to associate files with
actions on them.  This file is kept simple so the the users can edit it.
There is no way I can implement parsing of the BROWSER variable in one
line.  Either it needs to be done in the C code or there should be a shell
wrapper that would parse $BROWSER.

> But this X:noX syntax is not required: if KDE define BROWSER=konqueror
> at init, it's ok, X is started.
>
> So the ability to define multiple browser do not seems to me to be a
> top requirement. If we keep the request simple, we have better chance
> to see it implemented.

Agreed 100%.

> What do you think Pavel? If you agree, I'll ask KDE people to see if
> they are willing to implement a simple BROWSER putenv() at kdeinit.

Yes.   Make sure to tell them that simple BROWSER=konqueror would still
comply with the full standard, despite its complexity.

-- 
Regards,
Pavel Roskin