From: Curt Hibbs <curt@...>
Date: 2001-12-08T05:03:02+09:00
Subject: [ruby-talk:27846] Re: New RubyGarden Poll - this one affectsus all :)

I agree, I find myself using google groups more and more.

Curt

-----Original Message-----
From: Conrad Schneiker [mailto:schneiker@jump.net]
Sent: Friday, December 07, 2001 11:55 AM
To: ruby-talk ML; undisclosed-recipients:
Subject: [ruby-talk:27844] Re: New RubyGarden Poll - this one affectsus
all :)


Hi,

"Matt Armstrong" <matt+dated+1008173266.149069@lickey.com> wrote :

> "Conrad Schneiker" <schneiker@jump.net> writes:
[...]
> > Notably, "Dr. Dobb's Python-URL! - weekly Python news and links...."
(posted
> > weekly to comp.lang.python) uses references such as
> > "http://groups.google.com/groups?th=18e6d8363f51b0bc,31" to refer to
past
> > posts. This is notable since others were discussing emulating this cool
> > weekly news summary service for Ruby. IIRC, like comp.lang.ruby,
> > comp.lang.python is also mirrored to a corresponding mail list.
>
> This is a bad idea, as the google.com archive could go away at any
> minute, invalidating references to all the old posts.
>
> As you quoted in your post (which I have not included here) this
> happened when the old deja.com archive went down.  I am quite annoyed
> when I find something on the net that claims to explain exactly what I
> want to know, yet it is full of old deja.com URLs that no longer work.
>
> It is MUCH better to have the article itself be tagged with a unique
> ID somehow by the mailing list software.  This way, the message has
> the same unique ID no matter what archiving service is archiving the
> list.

Good points, but I am still not really persuaded, in part due to some
further considerations.

For example, groups.google seem to be a suitable informal standard that many
people in other newsgroups are satisfied with (notwithstanding the
admittedly inconvenient demise of deja), and unlike other dot.com ventures,
Google doesn't really seem likely to go away any minute. Also, for some or
for many of us, groups.google seems faster and more convenient than looking
up stuff on the ML page, and so is more handy for making back-references.
Finally, in contrast to a URL, [ruby-talk:12345] references require an extra
couple of (seemingly typically slow) steps to look up, which I personally
find mildly obnoxious.

Maybe we need another vote here. :-)

Conrad