Re: patch for 0.6.0.20070501-2:frontpage_controller

On 5/5/07, boud <boudriseup.net> wrote:
> i didn't mean to auto-generate the config file in the
debian package.
> i rather meant just to give a warning to the sysadmin
who is updating
> interactively. The only option s/he will have is to
click OK ;) - i'm
> not sure offhand which packages have this type of
messages, but i know
> that some do - it can even be: "Warning: there are
some important changes
> in version 0.6.0-20070501 - your installation will
probably fail unless
> you follow the recommendations in
/usr/share/doc/samizdat/NEWS.Debian."

Ah, you're right, I should put a warning using debconf.
Probably not
as critical for experimental, but definitely should be done
before
uploading this version to unstable.

> > automated way will always work. And yes, you are
supposed to read
> > NEWS.Debian when upgrading a package, even if it
doesn't break
> > immediately after upgrade.
> Even when an upgrade gives 10-20 packages modified? 
Well, i guess
> there's a difference between stable and
testing/unstable and experimental.

Don't take my word on this, some DDs may disagree with me


I am upgrading from unstable daily, so I'm kind of used to
checking
NEWS and changelogs when something breaks. Although I have
to admin it
doesn't happen all that often, and it's not supposed to
happen at all
when you upgrade from stable.

> On a related topic, what are the plans for getting
samizdat from unstable
> into the full unstable -> testing cycle, so that in
18-24 months' time
> samizdat will get into stable?

Version 0.6.0 is already in testing:
http://
packages.qa.debian.org/s/samizdat.html

I hope that in 18-24 months time it will be at least 0.6.1
there 

> >> BUG: language_list was missing :(((, and
language_list gave an old style
> >> URL member.rb?set_lang= which fails

Version 0.6.0.20070506-1 (soon to appear in experimental)
will include
this and a bunch of other fixes, including weird database
deadlocks
caused by CSRF protection. I hope the fix to keep action
tokens in
cache instead of database will not cause cache deadlocks
instead 

-- 
Dmitry Borodaenko


_______________________________________________
samizdat-devel mailing list
samizdat-develnongnu.org
http://lists.nongnu.org/mailman/listinfo/samizdat-devel

hi Dmitry,

On Sun, 6 May 2007, Dmitry Borodaenko wrote:

> On 5/5/07, boud <boudriseup.net> wrote:
>> i didn't mean to auto-generate the config file in
the debian package.
>> i rather meant just to give a warning to the
sysadmin who is updating
>> interactively. The only option s/he will have is to
click OK ;) - i'm
>> not sure offhand which packages have this type of
messages, but i know
>> that some do - it can even be: "Warning: there
are some important changes
>> in version 0.6.0-20070501 - your installation will
probably fail unless
>> you follow the recommendations in
/usr/share/doc/samizdat/NEWS.Debian."
>
> Ah, you're right, I should put a warning using debconf.
Probably not
> as critical for experimental, but definitely should be
done before
> uploading this version to unstable.

OK. i guess even in testing/unstable, sysadmins should
expect stuff to
break, i was thinking more in terms of the stable
distribution - and i 
guess in stable, essentially only security updates get
accepted, so this
sort of radical change would not go directly into stable.

>> > automated way will always work. And yes, you
are supposed to read
>> > NEWS.Debian when upgrading a package, even if
it doesn't break
>> > immediately after upgrade.
>> Even when an upgrade gives 10-20 packages modified?
 Well, i guess
>> there's a difference between stable and
testing/unstable and experimental.
>
> Don't take my word on this, some DDs may disagree with
me 



> I am upgrading from unstable daily, so I'm kind of used
to checking
> NEWS and changelogs when something breaks. Although I
have to admin it
> doesn't happen all that often, and it's not supposed to
happen at all
> when you upgrade from stable.

OK, sounds reasonable to me.

>> On a related topic, what are the plans for getting
samizdat from unstable
>> into the full unstable -> testing cycle, so that
in 18-24 months' time
>> samizdat will get into stable?
>
> Version 0.6.0 is already in testing:
> http://
packages.qa.debian.org/s/samizdat.html

Ah yes, you're right.

> I hope that in 18-24 months time it will be at least
0.6.1 there 

i would hope 0.6.2 

>> >> BUG: language_list was missing :(((, and
language_list gave an old style
>> >> URL member.rb?set_lang= which fails
>
> Version 0.6.0.20070506-1 (soon to appear in
experimental) will include
> this and a bunch of other fixes, including weird
database deadlocks
> caused by CSRF protection. I hope the fix to keep
action tokens in
> cache instead of database will not cause cache
deadlocks instead 

As i said in my email a moment ago about the calendar-0.4
patch, i think
i experienced a bug which was probably this "database
deadlock", and it
seems to have disappeared in 20070506. 


BTW - could we get a fix for the "upload content from
file" usability
issue into cvs?

We discussed this in February:

user barrier - ambiguity about "content" vs
"upload content from file"
http://lists.gnu.org/archive/html/samizdat
-devel/2007-02/msg00004.html
http://lists.gnu.org/archive/html/samizdat
-devel/2007-02/msg00005.html
http://lists.gnu.org/archive/html/samizdat
-devel/2007-02/msg00009.html

and i think it should be fairly easy.


cheers
boud


_______________________________________________
samizdat-devel mailing list
samizdat-develnongnu.org
http://lists.nongnu.org/mailman/listinfo/samizdat-devel