Re: New snapshot: candidate for 0.6.0

[Dmitry Borodaenko]

On 10/26/06, boud <address@hidden> wrote:
> Package samizdat
>
>      * experimental (web): Collaboration and open publishing engine
>        0.5.5.20060924-1: all
>
> It looks like it's still the 20060924 version to me - maybe the mirrors
> need time to copy? Or there's some sort of authentification process which
> takes a day or two?

There are many delays involved: it takes some time for a package to
get through incoming queue and hit the master archive (usually less
than a day, but sometimes more, especially for new packages (as
opposed by new versions of old packages)), then it takes another day
to propagate through mirrors (none are updated more than once a day),
and the packages web pages are updated even less often.

Use this link:
http://qa.debian.org/developer.php?login=angdraug

It reflects new package as soon as it hits the incoming queue.

> BTW, what hash function is used for the .sig signatures here?
> http://download.savannah.gnu.org/releases/samizdat/
>
> It doesn't seem to be either md5sum or sha1sum...

It is a detached PGP signature, check it against my public key (in
Debian keyring). This way, you not only can check the integrity of the
file, but verify that it actually comes from me (that is, if my key is
can be reached by your PGP web of trust).

> How about at least turning on the \n split for html? It's only a one
> line change, it's hard to see any problems it could cause, and it could
> reduce the number of users who publish articles in html and then get
> discouraged by an apparently poorly functioning diff engine.
>
> What i mean is just (1) from:
> http://lists.gnu.org/archive/html/samizdat-devel/2006-10/msg00007.html

Right, thanks for reminding of this one. Try this patch:

@@ -607,11 +607,13 @@ version 2 or later. </div>}
      msg[:content] =
      case msg[:format]
      when nil
-        msg[:content].split("\n\n").collect {|s| s << "\n\n" }
+        msg[:content].split(/(?=\n\n)/)
      when 'text/plain', 'text/textile'
-        msg[:content].split("\n").collect {|s| s << "\n" }
+        msg[:content].split(/(?=\n)/)
+      when 'text/html'
+        msg[:content].split(%r{(?=<(?!\s*/))})
      else
-        [msg[:content]]   # todo: html diff
+        [msg[:content]]
      end
    end
    %{<table class="diff">\n<thead><tr>\n} +

It splits HTML by starting tags, which seems to work even better.

> Hmmm...  i just noticed that "calendar" is for version 0.8.0 according
> to TODO.  This is actually on the "what we have" list for the imc-cms,
> i guess maybe it's present in sf-active, but it's certainly not present
> in mir - which led many groups to use a non-free calendar package on
> another server...    i guess this means you (dmitry) probably have
> something quite clever and solid in mind for a calendar feature, but
> i might try my hand on at least something simple which can be used until
> something more clever comes along - IMHO this is a fairly important
> feature.

Yes, what I have in mind for 0.8.0 is a proper groupware, something
more simple might as well sneak into 0.6.x, especially if it's not me
who implements it :)

> i had a go at a graph layout visualisation module, but it's probably
> not top priority right now. i will show what it looks like and ask
> for feedback though. At the moment it works :) but is incredibly slow,
> possibly due to the fact that the ruby-graphviz package calls graphviz
> through system calls.  Anyway, i'll post that soon in an independent
> email.

Hm, last time I took a look at graphviz, it was producing
one-and-a-half dimensional graph layouts, not really helpful for
anything more than handful of nodes. How good is it now?

Did you look in the direction of interactive SVG?

--
Dmitry Borodaenko