pkg kde talk - Jul 2008 - Working with the tools - or around/against the tools

[Proposal to change how we do stuff]

Hi!

Currently, when working with the changelog, we seem to in the team to have our 
own specific style, a style that requires to adapt the changelog from the style 
generated with usually recommended tools like dch and uupdate.

Every time I work around a tool, I think something is wrong. Either in the 
tool or in my workflow.

I hereby suggest that we change our style of doing changelogs into the way dch 
does it.

Currently, we have a style like:

package (1.2.3-4) experimental; urgency=low

  +++ Changes by Joe User:

  * doing foo
  * doing bar

  +++ Changes by Jane Hacker:

  * Something third
  * Something fourth

  --  Team-name <mail-list at host> date

where dch implements it as

package (1.2.3-4) experimental; urgency=low

  [ Joe User ]
  * doing foo
  * doing bar

  [ Jane Hacker ]
  * Something third
  * Something fourth

  -- Name of last changer <name at server> date

and dch does all the formatting automatically, so when Jack Third needs to add 
a entry, he just do dch "Fix typo in package description"  and the
resulting
changelog entry says:

package (1.2.3-4) experimental; urgency=low

  [ Joe User ]
  * doing foo
  * doing bar

  [ Jane Hacker ]
  * Something third
  * Something fourth

  [ Jack Third ]
  * Fix typo in package description

  -- Jack Third <jthird at world> date

So please. Let us adopt the way dch does it so we don''t have to change
and
adapt the changelog all the time.  This also have the advantage that doing 
automatic processing, dch can do the changelog handling instead of own 
scripts.

Now speak up - and by the way, I will consider this as accepted if no one 
speaks against it within the next weeks or so.  If most people currently in 
uploaders fields of the packages accept it, doing it quicker is also possible.

/Sune
-- 
I cannot forward to the attachment, how does it work?

First you should cancel the mouse for removing a folder.

On Thursday 03 July 2008, Sune Vuorela wrote:
> So please. Let us adopt the way dch does it so we don''t have to
change and
> adapt the changelog all the time.  This also have the advantage that doing
> automatic processing, dch can do the changelog handling instead of own
> scripts.
We should at least move over to the "[ /me ]" format instead of
"++ /me", so
we don''t have to keep changing that.  We might as well adopt the
"standard"
behaviour if it doesn''t hurt.

What''s really missing here is the ability to *only sometimes* use
multimaint
mode in dch.  Currently, it''s either always-on or always-off, when what
would
be far more useful would be an option to enable it only for a few 
names/addresses specified in the config file.  That would be a wishlist bug 
against devscripts, and a much-needed one for any team-maintained package.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part.
Url :
http://lists.alioth.debian.org/pipermail/pkg-kde-talk/attachments/20080703/ad7f68ca/attachment.pgp

On Thu, Jul 03, 2008 at 06:04:36PM +0200, Sune Vuorela
wrote:
> [Proposal to change how we do stuff]
> 
> Hi!
> 
> Currently, when working with the changelog, we seem to in the team to have
our
> own specific style, a style that requires to adapt the changelog from the
style
> generated with usually recommended tools like dch and uupdate.
> 
> Every time I work around a tool, I think something is wrong. Either in the 
> tool or in my workflow.
> 
> I hereby suggest that we change our style of doing changelogs into the way
dch
> does it.
> 
> Currently, we have a style like:
> 
> package (1.2.3-4) experimental; urgency=low
> 
>   +++ Changes by Joe User:
> 
>   * doing foo
>   * doing bar
> 
>   +++ Changes by Jane Hacker:
> 
>   * Something third
>   * Something fourth
> 
>   --  Team-name <mail-list at host> date
> 
> where dch implements it as
> 
> package (1.2.3-4) experimental; urgency=low
> 
>   [ Joe User ]
>   * doing foo
>   * doing bar
> 
>   [ Jane Hacker ]
>   * Something third
>   * Something fourth
> 
>   -- Name of last changer <name at server> date
> 
> and dch does all the formatting automatically, so when Jack Third needs to
add
> a entry, he just do dch "Fix typo in package description"  and
the resulting
> changelog entry says:
> 
> package (1.2.3-4) experimental; urgency=low
> 
>   [ Joe User ]
>   * doing foo
>   * doing bar
> 
>   [ Jane Hacker ]
>   * Something third
>   * Something fourth
> 
>   [ Jack Third ]
>   * Fix typo in package description
> 
>   -- Jack Third <jthird at world> date
> 
> So please. Let us adopt the way dch does it so we don''t have to
change and
> adapt the changelog all the time.  This also have the advantage that doing 
> automatic processing, dch can do the changelog handling instead of own 
> scripts.
>
I do not care at all about using +++ Name or [ Name ] but when more than a
person have worked in the package I like uploading it as a team upload as we
have done until now. So not ok about changing that.
> Now speak up - and by the way, I will consider this as accepted if no one 
> speaks against it within the next weeks or so.  
This is not a nice way of handling stuff... so I send a mail with big changes
the week you are off on holidays with the same way of "agreement" and
it is ok?
Sometimes mail with proposal do not have answer because well... it is not
worth answer it.
> If most people currently in 
> uploaders fields of the packages accept it, doing it quicker is also
possible.
> 
Why? We have plenty of people in the Uploaders field and we all know not
everybody works the same in the team. We should try to reach concensus using
the common sense, not a majority vote that does not make sense when work is
not balanced.


Ana

On Thursday 03 July 2008 20:46:00 Ana Guerrero wrote:
> > Now speak up - and by the way, I will consider this as accepted if no
one
> > speaks against it within the next weeks or so.
>
> This is not a nice way of handling stuff... so I send a mail with big
> changes the week you are off on holidays with the same way of
"agreement"
> and it is ok? 
No one has announced being on vacation - and if people leave without a 
announcement of vacation, we can''t wait forever on them.
> We have plenty of people in the Uploaders field and we all know not
> everybody works the same in the team. We should try to reach concensus
> using the common sense, not a majority vote that does not make sense when
> work is not balanced.
Do I read this as you saying "I think I do most work, so I decide"
correctly ?

And if it was about reaching consensus, we would still be discussing kde3 vs 
kde4.

/Sune
-- 
Man, do you know how could I send the terminale from the control preferences 
inside Excel?

You cannot connect the LCD IRC prompt.

Hi,

Thursday 03 July 2008, Ana Guerrero ra??:
> I do not care at all about using +++ Name or [ Name ] but when more than a
> person have worked in the package I like uploading it as a team upload as
> we have done until now. So not ok about changing that.
However, current practise is not consistent with regard to if there is only 
one maintainer doing changes, team name & address may not be used at the 
bottom. This practise should be avoided because:

1) Somebody is _always_ the first to make changes. So s(he) may put his/her 
name in the maintainer space of the changelog entry (lets call such entry 
entry).
2) Now the second maintainer committing changes has to:
  a) add +++ Changes by line for the first maintainer, which effectively means 
copy&pasting from the changelog maintainer field if 1st maitainer created a 
personal entry. A bit time consuming.
  b) add "+++ Changes for My name". Pretty easy.
  c) replace current changelog maintainer with Debian QT/KDE Maintainers. I 
usually end up copying and pasting from the previous non-personal entry (but I 
sometimes need to search for the it due to existence of personal entries).
3) Third and subsequent maintainners has only to do 2b. So the 1st or the 2nd 
maintainner gets the most additional work to do with the changelog.
4) If personal changelog entries are allowed, it effectively means that dch -m 
will not work for creating proper new revision entry after personal changelog 
entry. So the 1st maintainer has to do 2c in addition to 2b or create a new  
personal entry _or_ leave all the burden to the 2nd maintainer.

Most of these tasks can be automated by dch if the standard layout is used. 
However, dch behaviour depends on having the name of the real maintainer in 
the changelog maintainer field to transform personal entry to the team entry.

Having all this said, I very dislike personal changelog entries because they 
potentially leave more work for the 2nd maintainer (2a part). I never do them. 
So if to change anything at all, in my opinion, only the following 
alternatives are viable:

1) Do as dch supports now, i.e. do not use team in the changelog maintainer 
field (Sune proposal).
2) Do not change anything until "team name in the changelog
maintainer" is
implemented in dch (Ana''s wish).
3) Disallow non-team name & address in the changelog mainainer field, hence 
everybody adds their own "+++ Changes by name". I can write svn hook
which
would enforce this rule. This will ensure that dch -m will always work 
properly for creating new changelog entries.

-- 
Modestas Vainius <modestas at vainius.eu>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part.
Url :
http://lists.alioth.debian.org/pipermail/pkg-kde-talk/attachments/20080703/912212ec/attachment.pgp

Hi,

Alle gioved? 03 luglio 2008, Modestas Vainius ha
scritto:
> Thursday 03 July 2008, Ana Guerrero ra??:
> > I do not care at all about using +++ Name or [ Name ] but when more
than
> > a person have worked in the package I like uploading it as a team
upload
> > as we have done until now. So not ok about changing that.
Fully agree with this; as long as there''s something
"coherent" with the
majority of ther packages in Debian, that''s fine.
> 2) Now the second maintainer committing changes has to:
>   a) add +++ Changes by line for the first maintainer, which effectively
> means copy&pasting from the changelog maintainer field if 1st maitainer
> created a personal entry. A bit time consuming.
>   b) add "+++ Changes for My name". Pretty easy.
>   c) replace current changelog maintainer with Debian QT/KDE Maintainers.
Another (minor) disadvantage of changing "non-named" entries to
"named" ones
is reindenting the entries: not for the "hassle" of doing that, but
also for
the slightly "tampering" of the VCS "blame" feature.
> 2) Do not change anything until "team name in the changelog
maintainer" is
> implemented in dch (Ana''s wish).
My vote would go to this...
> 3) Disallow non-team name & address in the changelog mainainer field,
hence
> everybody adds their own "+++ Changes by name". I can write svn
hook which
> would enforce this rule. This will ensure that dch -m will always work
> properly for creating new changelog entries.
... but this proposal is basically b), just done manually, so +1 on this, too.
And, as long as a dch-like format is used, this should not be incompatible 
with b), I think.

We are a team, so let''s act as team! :)

-- 
Pino Toscano
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part.
Url :
http://lists.alioth.debian.org/pipermail/pkg-kde-talk/attachments/20080703/7adea6f3/attachment.pgp

On Thu, Jul 03, 2008 at 09:12:18PM +0200, Sune Vuorela
wrote:
> 
> On Thursday 03 July 2008 20:46:00 Ana Guerrero wrote:
> 
> > > Now speak up - and by the way, I will consider this as accepted
if no one
> > > speaks against it within the next weeks or so.
> >
> > This is not a nice way of handling stuff... so I send a mail with big
> > changes the week you are off on holidays with the same way of
"agreement"
> > and it is ok? 
> 
> No one has announced being on vacation - and if people leave without a 
> announcement of vacation, we can''t wait forever on them.
> 
> > We have plenty of people in the Uploaders field and we all know not
> > everybody works the same in the team. We should try to reach concensus
> > using the common sense, not a majority vote that does not make sense
when
> > work is not balanced.
> 
> Do I read this as you saying "I think I do most work, so I
decide" correctly ?
>
As usual, you read what you want to read and attack with wathever excuse you
have. You have people in the team-members file with have worked a lot in the 
pass and is not currently active (schepler), you have people who is limited to 
a set of stuff (dato), you have people who has barely started working with us
(pino), you have people who did 2-3 commits and not longer contributed (Sara) 
and so so. The set of the "usual workers" is 4-5 people and you have
17 listed
in members-team.  

> And if it was about reaching consensus, we would still be discussing kde3
vs
> kde4.
Concesus is about sometimes have to give in something realising you may be
wrong even if you do not see it.

On Thursday 03 July 2008, Modestas Vainius wrote:
> Most of these tasks can be automated by dch if the standard layout is used.
> However, dch behaviour depends on having the name of the real maintainer in
> the changelog maintainer field to transform personal entry to the team
> entry.
So what is our wishlist for dch?

I was thinking perhaps that a few team names/addresses could be specified in 
the devscripts conf.  dch would always use multimaint mode for those addresses 
and preserve the team address at the bottom of the changelog entry.

Any other ideas, or should I file a wishlist bug for this behaviour?
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part.
Url :
http://lists.alioth.debian.org/pipermail/pkg-kde-talk/attachments/20080703/5b0ccf83/attachment-0001.pgp

Hi,

Friday 04 July 2008, Matthew Rosewarne ra??:
> I was thinking perhaps that a few team names/addresses could be specified
> in the devscripts conf.  dch would always use multimaint mode for those
> addresses and preserve the team address at the bottom of the changelog
> entry.
>
> Any other ideas, or should I file a wishlist bug for this behaviour?
I would suggest a switch for dch to always enforce multimaitainer mode and use 
debian/control Maintainer field value as the changelog maintainer.

-- 
Modestas Vainius <modestas at vainius.eu>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part.
Url :
http://lists.alioth.debian.org/pipermail/pkg-kde-talk/attachments/20080704/a1a0fa9b/attachment.pgp

Hi,

IMHO, I prefer dch multimaintainer mode implementation.

by the way, we can start to use this scheme manually, no need to wait for dch 
changes.

package (1.2.3-4) experimental; urgency=low

  [ Joe User ]
  * doing foo
  * doing bar

  [ Jane Hacker ]
  * Something third
  * Something fourth

  --  Team-name <mail-list at host> date

cheers,

Fathi