Re: [RFR] templates://webfs/{templates}
tisdag den 2 mars 2010 klockan 07:40 skrev Christian PERRIER detta:
> Quoting Mats Erik Andersson (mats.andersson@gisladisker.se):
> > Hello,
> >
>
> I made most of your suggested corrections.
>
> Except:
>
> > Template: webfsd/web_root
> > Type: string
> > -Default: /var/ftp
> > +Default: /home/ftp
>
>
> Is this something that you would use, as maintainer, as an opportunity
> to change the package behaviour?
Yes. The value '/var/ftp/' breaks FHS policy and ought not be commended.
I already changed from '/www/ftp' as my first action when adopting the package,
and I have after thinking it through, found that '/home/ftp/' would be a
more plausible value than '/srv/ftp/' for small systems, the intended
audience. I do not any longer remember how it was suggested, or came about,
but the two most recent times I set 'proftpd-basic' up for smaller work,
the choice '/home/ftp/' came out of it, and it works very well.
I still expect many, if not most, users to change this field. The really
important matter is to get away from '/www/ftp/' as a standard answer.
However, after reading FHS yet another time, and in spite of what I just
wrote, I concede '/srv/ftp/' to be the default string of choice. A consistent
use of FHS is better that an effort to guess common choices of users.
>
> > _Description: Document root for webfsd:
> > Webfsd is a lightweight HTTP server for mostly static content. Its
> > most obvious use is to provide HTTP access to an anonymous FTP server.
> > @@ -55,7 +56,7 @@
> > _Description: Host name for webfsd:
> > By default, webfsd uses the machine name as hostname.
> > .
> > - You can specify an alternate host name to be used as an external
> > + You can specify an alternate hostname to be used as an external
> > alias name (for instance "ftp.example.org") instead of the machine's
> > fully qualified domain name.
> >
> > Consistency in writing 'hostname'.
>
> In such case, it should be "host name". And the "by default" sentence
> should be changed too.
>
Yes, also I understand 'host name' as the preferred orthography.
>
> > As an experienced TeX and LaTeX user, I am conscious of the notion of
> > French spacing. The continental Western European and Northern European
> > languages are using a single space after a full period, whereas the
> > North American habit, as well as the the Anglosaxon habit, as far as I
> > understand, is to use double spacing after a full stop. Anyone can
>
> This is only an habit and apparently less and less consistently
> developed, including North America, not to mention UK which apparently
> never used it. This is indeed mostly an heritage of the typewriters days
> and monospaced fonts.
>
> Indeed, most typographical recommendations now seem to agree on 1 em
> spacing after full stops....and that a single space should be used in
> electronic material
> (http://en.wikipedia.org/wiki/Double_spacing_at_the_end_of_sentences
> gives enough references, including USA and Canada ones).
>
> The most important reference wrt this is the Chicago Manual of Style,
> which is commonly accepted as the best reference in USA for
> typographical conventions.
>
> After talking about this many times with northern American users, they
> apparently use double spacing after full stops in a quite
> inconsistent way: some do, some don't. After some thinking and even
> though we standardized on US-like spelling (even though most reviewers
> are en_GB natives), we decided to standardize to single space after
> full stops...mostly because this is what requires less changes that
> might be seen as useless.
>
> So, well, from all these references, we seem to have enough backup to
> sustain our fight against windmil^W double spaces..:-)
>
Agreed. So it is legitimately annoying that LaTeX uses double spacing
by default! I appreciate the extended explanation. It helps a lot!
Regarding the text template for choosing the listening port: I do not find
the setting of a default string to be optimal. The reason is that Webfds,
when not recording '-p num' will take 'num=8000' for implicit, whereas the
init.d script inserts '-p arg' only when the debconf item has non-void
content. Thus the setting of a default string would enforce a superfluous
command line option whenever the default choice is intended. This disrupts
somewhat the clarity of whatever non-trivial options the commandline really
did ask the service to grant.
Regards,
--
Mats Erik Andersson, fil. dr
<mats.andersson@gisladisker.se>
Abbonerar på: debian-mentors, debian-devel-games, debian-perl, debian-ipv6
Reply to: