[gentoo-dev] Planning for automatic assignment of bugs

[gentoo-dev] Planning for automatic assignment of bugs

[Robin H. Johnson]

[-- Attachment #1: Type: text/plain, Size: 2615 bytes --]

So as a not-so-brief follow-up to solar's email, here is a brief
proposal on the automatic assignment stuff, incl. one spot that we might
need to add an attribute to metadata.xml.

Assignment process, triggering:
===============================
Auto-assignment will be be applied/available in the following cases:
1. New bugs created with the guided process, having a Product equal to
   'Gentoo Linux' and a component not equal to 'Eclasses and Profiles'.
2. Open bugs will have a new action available: 'Reassign by metadata',
   with a text input field. The text field will be auto-filled with a
   package atom $CAT/$PN by parsing the summary line. Using the action
   will provide the package atom to the next stage.

If multiple package atoms are present in a summary line, the first one
wins.

Assignment process, after the package is known:
===============================================

We have a package spec now, so we can find who to assign the bug to.

Objectives in this section are to reduce unwanted duplicate mail, while
still preserving the data in metadata for non-automated usage.

Case 1 - Metadata contains only a herd
--------------------------------------
- The herd will have @gentoo.org appended, and this must be a valid
  bugzilla account.

Case 2 - Metadata contains a single maintainer
----------------------------------------------
- The herd field is not used.
- The maintainer address is used as the bugzilla assignee. 
This is important for all the herds that have aliases that are NOT the
same as their herd name!
This diverges from existing manual practice, to avoid unnecessary
duplicate mail, and means that existing metadata may need a cleanup.

Case 3 - Metadata contains multiple maintainers
-----------------------------------------------
- Follow case 2 first.
- Further maintainer addresses are used in the CC field.

Case 4 - Metadata contains multiple maintainers, some special
-------------------------------------------------------------
- Follow case 3 first.
- If a maintainer is listed in the metadata for special reasons (eg only
  for some special patch), they should include the 'contact=0' attribute
  on their maintainer element AND have a role element present
  describing why.
- This also allows for cases where the herd address should be used as
  the assignee, and the maintainer does NOT want a duplicate CC.

Comments etc welcome.

-- 
Robin Hugh Johnson
Gentoo Linux Developer & Council Member
E-Mail     : robbat2@gentoo.org
GnuPG FP   : 11AC BA4F 4778 E3F6 E4ED  F38E B27B 944E 3488 4E85

[-- Attachment #2: Type: application/pgp-signature, Size: 321 bytes --]

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Robin H. Johnson]

[-- Attachment #1: Type: text/plain, Size: 492 bytes --]

Correction to the previous email, this replaces the old Case #1:

On Thu, Apr 26, 2007 at 12:40:06PM -0700, Robin H. Johnson wrote:
> Case 1 - Metadata contains only a herd
> --------------------------------------
- The herd will looked up in herds.xml, and the email from herds.xml
  must be a valid bugzilla account.

-- 
Robin Hugh Johnson
Gentoo Linux Developer & Council Member
E-Mail     : robbat2@gentoo.org
GnuPG FP   : 11AC BA4F 4778 E3F6 E4ED  F38E B27B 944E 3488 4E85

[-- Attachment #2: Type: application/pgp-signature, Size: 321 bytes --]

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Dan Meltzer]

On Thursday 26 April 2007 3:40:06 pm Robin H. Johnson wrote:
> So as a not-so-brief follow-up to solar's email, here is a brief
> proposal on the automatic assignment stuff, incl. one spot that we might
> need to add an attribute to metadata.xml.
>
> Assignment process, triggering:
> ===============================
> Auto-assignment will be be applied/available in the following cases:
> 1. New bugs created with the guided process, having a Product equal to
>    'Gentoo Linux' and a component not equal to 'Eclasses and Profiles'.
> 2. Open bugs will have a new action available: 'Reassign by metadata',
>    with a text input field. The text field will be auto-filled with a
>    package atom $CAT/$PN by parsing the summary line. Using the action
>    will provide the package atom to the next stage.
>
> If multiple package atoms are present in a summary line, the first one
> wins.
>
> Assignment process, after the package is known:
> ===============================================
>
> We have a package spec now, so we can find who to assign the bug to.
>
> Objectives in this section are to reduce unwanted duplicate mail, while
> still preserving the data in metadata for non-automated usage.
>
> Case 1 - Metadata contains only a herd
> --------------------------------------
> - The herd will have @gentoo.org appended, and this must be a valid
>   bugzilla account.
>
> Case 2 - Metadata contains a single maintainer
> ----------------------------------------------
> - The herd field is not used.
> - The maintainer address is used as the bugzilla assignee.
> This is important for all the herds that have aliases that are NOT the
> same as their herd name!
> This diverges from existing manual practice, to avoid unnecessary
> duplicate mail, and means that existing metadata may need a cleanup.
>
> Case 3 - Metadata contains multiple maintainers
> -----------------------------------------------
> - Follow case 2 first.
> - Further maintainer addresses are used in the CC field.
>
> Case 4 - Metadata contains multiple maintainers, some special
> -------------------------------------------------------------
> - Follow case 3 first.
> - If a maintainer is listed in the metadata for special reasons (eg only
>   for some special patch), they should include the 'contact=0' attribute
>   on their maintainer element AND have a role element present
>   describing why.
> - This also allows for cases where the herd address should be used as
>   the assignee, and the maintainer does NOT want a duplicate CC.
>
> Comments etc welcome.

Sounds good... one suggestion I have is to try and detect new ebuild 
submissions and resassign them to m-w automatically as well.  maybe a 
checkbox "this is a new ebuild" or some other way to automatically detect it?
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[expose]

Dan Meltzer wrote:
> Sounds good... one suggestion I have is to try and detect new ebuild
> submissions and resassign them to m-w automatically as well.  maybe a
> checkbox "this is a new ebuild" or some other way to automatically detect
> it?
Why not introduce a "Case 5" which is similar to:
1. None of the other Cases worked
2. There is a package atom in the summary line
(2a. Words like "ebuild" "new" occur somewhere)
3. Check wether the package matches a "class of packages"
4. If it doesnt match any classes, leave it to the wranglers.

package classes would be something like:
If the package is of the games-* category assign it to the games herd.
If the package matches */*python* assign it to $whoever

Unless the specifications for a package were not met correctly, or the 
submitter did not include a atom in the summery as expected, the checkbox 
shouldn't be needed, because the bug should be assigned by either one of 
cases 1 to 4. If it isnt, case 5 (or whatever else automatic handling for 
new-ebuilds) should be suiteable for trying to fix it anyway.

Comments?
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Nguyen Thai Ngoc Duy]

On 4/26/07, Robin H. Johnson <robbat2@gentoo.org> wrote:
> Case 2 - Metadata contains a single maintainer
> ----------------------------------------------
> - The herd field is not used.
> - The maintainer address is used as the bugzilla assignee.
> This is important for all the herds that have aliases that are NOT the
> same as their herd name!
> This diverges from existing manual practice, to avoid unnecessary
> duplicate mail, and means that existing metadata may need a cleanup.

It should take devaway into account.

-- 
Duy
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Joshua Jackson]

[-- Attachment #1: Type: text/plain, Size: 890 bytes --]

Nguyen Thai Ngoc Duy wrote:
> On 4/26/07, Robin H. Johnson <robbat2@gentoo.org> wrote:
>> Case 2 - Metadata contains a single maintainer
>> ----------------------------------------------
>> - The herd field is not used.
>> - The maintainer address is used as the bugzilla assignee.
>> This is important for all the herds that have aliases that are NOT the
>> same as their herd name!
>> This diverges from existing manual practice, to avoid unnecessary
>> duplicate mail, and means that existing metadata may need a cleanup.
>
> It should take devaway into account.
>
why? Seriously, dev-away != dev retired... having it take devaway into
account is pointless in my opinion as it won't improve it being properly
assigned...as it'll be covered in other cases, and its not like there's
not bugs for all of us dev's that have not sat there for a month or so,
at some point


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 252 bytes --]

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Daniel Drake]

Robin H. Johnson wrote:
> Case 2 - Metadata contains a single maintainer
> ----------------------------------------------
> - The herd field is not used.
> - The maintainer address is used as the bugzilla assignee. 

At least for some packages I'm involved with, this will result in me 
deleting myself from metadata.xml (but I'd rather not do so).

I like these bugs to go to the herd, not me directly. I get the bug mail 
anyway (I'm in the herd) but sometimes other herd members who see the 
mail jump in and help resolve the bug, for which I'm very grateful.

That aside, I like having myself in the metadata alongside the herd, to 
point out that I am the primary maintainer within the herd for the 
package in question. It is also useful for others so that when they have 
questions about the package, they know who to approach on IRC or whatever.

Apart from those small concerns, I like the idea of what you are proposing.

Thanks,
Daniel

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[expose]

Joshua Jackson wrote:
> Nguyen Thai Ngoc Duy wrote:
> > On 4/26/07, Robin H. Johnson <robbat2@gentoo.org> wrote:
> >> Case 2 - Metadata contains a single maintainer
> >> ----------------------------------------------
> >> - The herd field is not used.
> >> - The maintainer address is used as the bugzilla assignee.
> >> This is important for all the herds that have aliases that are NOT the
> >> same as their herd name!
> >> This diverges from existing manual practice, to avoid unnecessary
> >> duplicate mail, and means that existing metadata may need a cleanup.
> >
> > It should take devaway into account.
>
> why? Seriously, dev-away != dev retired... having it take devaway into
> account is pointless in my opinion as it won't improve it being properly
> assigned...as it'll be covered in other cases, and its not like there's
> not bugs for all of us dev's that have not sat there for a month or so,
> at some point
I think it is important that someone has a look at the bugs submitted.
For instance, a submitter might not behave correctly, and submit a security 
issue to Gentoo Linux instead of Gentoo Security. Or the submitter does not 
even recognize it is a security issue, or might be or become one.
In that case the bug would sit there until the dev it was automatically 
assigned to is back again (which could take quite some time).

Another (minor) reason, might be, that bugs could be assigned to the wrong 
person, because of whatsoever reason, and that dev is unavailable.
This bug would sit there and wait to be reassigned until the dev comes back, 
who will reassign it to the correct dev, who in turn will mark it 
as "dublicate" because someone else useing the correct format for the summary 
submitted it yet and the submitter of our first bug feels demotivated :-)

Maybe it is for those and similar reasons which can easily be made up better 
to leave bugs which would otherwise be automatically assigned to devs which 
are unavailable to the bug-wranglers, or maybe better:
Automatically assign but manually review them to emulate the first glance the 
unavailable dev would have taken.

Obviously, there should not be too many bugs which would be automatically 
assigned to developers which are away anyway.
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[expose]

Daniel Drake wrote:
> That aside, I like having myself in the metadata alongside the herd, to
> point out that I am the primary maintainer within the herd for the
> package in question. It is also useful for others so that when they have
> questions about the package, they know who to approach on IRC or whatever.
As far as I understand it, wouldnt making you the first maintainer in the list 
but adding 'contact=0' and a comment that you are the primary maintainer 
within the herd, and then following the herd as the maintainer the bug is to 
be actually assigned to fix this?
Naturally most users will read your comment first as it is at the top of the 
file, thus know that you are the one to be contacted in case of random 
questions, but the herd will still be the bugzilla account the bug is 
assigned to.
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Robin H. Johnson]

[-- Attachment #1: Type: text/plain, Size: 1560 bytes --]

On Thu, Apr 26, 2007 at 05:46:24PM -0400, Daniel Drake wrote:
>  Robin H. Johnson wrote:
> > Case 2 - Metadata contains a single maintainer
> > ----------------------------------------------
> > - The herd field is not used.
> > - The maintainer address is used as the bugzilla assignee. 
>  At least for some packages I'm involved with, this will result in me 
>  deleting myself from metadata.xml (but I'd rather not do so).
> 
>  I like these bugs to go to the herd, not me directly. I get the bug mail 
>  anyway (I'm in the herd) but sometimes other herd members who see the mail 
>  jump in and help resolve the bug, for which I'm very grateful.
This is handled by a later case in the proposal.
Simply interest a maintainer element with the herd email address, and
add the contact=0 attribute to your maintainer element in the file.

>  That aside, I like having myself in the metadata alongside the herd, to 
>  point out that I am the primary maintainer within the herd for the package 
>  in question. It is also useful for others so that when they have questions 
>  about the package, they know who to approach on IRC or whatever.
This is exactly the reason that I proposed the contact=0 attribute - for
some of the packages that I maintain, I do not want the bugs assigned
directly to me, but to the herd instead. While for others I _do_ want
the duplicate.

-- 
Robin Hugh Johnson
Gentoo Linux Developer & Council Member
E-Mail     : robbat2@gentoo.org
GnuPG FP   : 11AC BA4F 4778 E3F6 E4ED  F38E B27B 944E 3488 4E85

[-- Attachment #2: Type: application/pgp-signature, Size: 321 bytes --]

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Mart Raudsepp]

On Thu, 2007-04-26 at 15:16 -0700, Robin H. Johnson wrote:
> On Thu, Apr 26, 2007 at 05:46:24PM -0400, Daniel Drake wrote:
> >  Robin H. Johnson wrote:
> > > Case 2 - Metadata contains a single maintainer
> > > ----------------------------------------------
> > > - The herd field is not used.
> > > - The maintainer address is used as the bugzilla assignee. 
> >  At least for some packages I'm involved with, this will result in me 
> >  deleting myself from metadata.xml (but I'd rather not do so).
> > 
> >  I like these bugs to go to the herd, not me directly. I get the bug mail 
> >  anyway (I'm in the herd) but sometimes other herd members who see the mail 
> >  jump in and help resolve the bug, for which I'm very grateful.
> This is handled by a later case in the proposal.
> Simply interest a maintainer element with the herd email address, and
> add the contact=0 attribute to your maintainer element in the file.
> 
> >  That aside, I like having myself in the metadata alongside the herd, to 
> >  point out that I am the primary maintainer within the herd for the package 
> >  in question. It is also useful for others so that when they have questions 
> >  about the package, they know who to approach on IRC or whatever.
> This is exactly the reason that I proposed the contact=0 attribute - for
> some of the packages that I maintain, I do not want the bugs assigned
> directly to me, but to the herd instead. While for others I _do_ want
> the duplicate.

Could "contact" be named differently then?

contact=0 in metadata.xml in this context means that the automatic
reassigning should not assign to that maintainer, but when a user looks
whom to ask specific questions from and sees contact=0 he/she will
understand he/she is not to contact that person as the value is zero,
but Daniel wants them to contact precisely him in that case.
A different keyword might be better for that reason.

Good proposal otherwise!
I do have some reservations due to no human looking over new bugs
(before they get reassigned to a possibly otherwise busy maintainer), as
someone already has expressed, but we can always try it out and see how
it goes, I think.

Regards,
Mart Raudsepp

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Robin H. Johnson]

[-- Attachment #1: Type: text/plain, Size: 970 bytes --]

On Fri, Apr 27, 2007 at 02:57:59AM +0300, Mart Raudsepp wrote:
> > This is exactly the reason that I proposed the contact=0 attribute - for
> > some of the packages that I maintain, I do not want the bugs assigned
> > directly to me, but to the herd instead. While for others I _do_ want
> > the duplicate.
> Could "contact" be named differently then?
'autocontact' then?
Both 'assign' and 'cc' (and derivations thereof are not suitable).

> I do have some reservations due to no human looking over new bugs
> (before they get reassigned to a possibly otherwise busy maintainer), as
> someone already has expressed, but we can always try it out and see how
> it goes, I think.
This happens already, simply observe the bugs that pile up for
understaffed herds, or developers that are AWOL.

-- 
Robin Hugh Johnson
Gentoo Linux Developer & Council Member
E-Mail     : robbat2@gentoo.org
GnuPG FP   : 11AC BA4F 4778 E3F6 E4ED  F38E B27B 944E 3488 4E85

[-- Attachment #2: Type: application/pgp-signature, Size: 321 bytes --]

Re: [gentoo-dev] Planning for automatic assignment of bugs

[expose]

Mart Raudsepp wrote:
> Could "contact" be named differently then?
>
> contact=0 in metadata.xml in this context means that the automatic
> reassigning should not assign to that maintainer, but when a user looks
> whom to ask specific questions from and sees contact=0 he/she will
> understand he/she is not to contact that person as the value is zero,
> but Daniel wants them to contact precisely him in that case.
> A different keyword might be better for that reason.
Good point.
AutoAssign=FALSE
NoAutoAssignment=TRUE
etc might be better.
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[expose]

Robin H. Johnson wrote:
> Both 'assign' and 'cc' (and derivations thereof are not suitable).
Why not?
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Danny van Dyk]

Am Freitag, 27. April 2007 schrieb Robin H. Johnson:
> On Fri, Apr 27, 2007 at 02:57:59AM +0300, Mart Raudsepp wrote:
> > > This is exactly the reason that I proposed the contact=0
> > > attribute - for some of the packages that I maintain, I do not
> > > want the bugs assigned directly to me, but to the herd instead.
> > > While for others I _do_ want the duplicate.
> >
> > Could "contact" be named differently then?
>
> 'autocontact' then?
> Both 'assign' and 'cc' (and derivations thereof are not suitable).
notification=assignment|cc|none ?

Danny
-- 
Danny van Dyk <kugelfang@gentoo.org>
Gentoo/AMD64 Project, Gentoo Scientific Project
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Andrej Kacian]

On Thu, 26 Apr 2007 17:24:06 -0700
"Robin H. Johnson" <robbat2@gentoo.org> wrote:

> On Fri, Apr 27, 2007 at 02:57:59AM +0300, Mart Raudsepp wrote:
> > > This is exactly the reason that I proposed the contact=0 attribute - for
> > > some of the packages that I maintain, I do not want the bugs assigned
> > > directly to me, but to the herd instead. While for others I _do_ want
> > > the duplicate.  
> > Could "contact" be named differently then?  
> 'autocontact' then?
> Both 'assign' and 'cc' (and derivations thereof are not suitable).

How about 'possessive'? :)

-- 
Andrej

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Robin H. Johnson]

[-- Attachment #1: Type: text/plain, Size: 1203 bytes --]

On Fri, Apr 27, 2007 at 02:33:50AM +0200, Danny van Dyk wrote:
> > Both 'assign' and 'cc' (and derivations thereof are not suitable).
> notification=assignment|cc|none ?
This is to answer expose's question as well, but the attribute should
only indicate if the maintainer entry should be used for any automatic
process at all, not how to use it.

One of the reasons is that multiple maintainers each with
notification=assignment obviously won't work, and it's non-trivial to
validate via the DTD (yes, DTDs suck compared to XSchema, I know).

I intend that the first non-excluded maintainer entry is the one used
for the automatic process.

In terms of implementing this in the DTD, I'm going to specify that
'contact=1' (or whatever name we settle on) is the default, so that we
don't break validation of any existing metadata:

<!ATTLIST maintainer
  contact   (0|1)   1   -- should this maintainer be used by 
                        -- automatic processes?
 >

In light of the above, how about 'automatic=0'?

-- 
Robin Hugh Johnson
Gentoo Linux Developer & Council Member
E-Mail     : robbat2@gentoo.org
GnuPG FP   : 11AC BA4F 4778 E3F6 E4ED  F38E B27B 944E 3488 4E85

[-- Attachment #2: Type: application/pgp-signature, Size: 321 bytes --]

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Jan Kundrát]

[-- Attachment #1: Type: text/plain, Size: 740 bytes --]

Robin H. Johnson wrote:
> In terms of implementing this in the DTD, I'm going to specify that
> 'contact=1' (or whatever name we settle on) is the default, so that we
> don't break validation of any existing metadata:
> 
> <!ATTLIST maintainer
>   contact   (0|1)   1   -- should this maintainer be used by 
>                         -- automatic processes?
>  >
> 
> In light of the above, how about 'automatic=0'? 

AFAIK the preferred way of specifying boolean values in XML is to use
contact="contact", not contact="1".

Speaking of name, I'd try to include string "bug" somewhere in the
attribute name, like "bugzilla-auto-assignment" or something...

Cheers,
-jkt

-- 
cd /local/pub && more beer > /dev/mouth


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 252 bytes --]

Re: [gentoo-dev] Planning for automatic assignment of bugs

[expose]

Robin H. Johnson wrote:
> [...] the attribute should only indicate if the maintainer entry should be
> used for any automatic process at all, not how to use it. 
Oh, I thought you were talking about the name of the variable.

> I intend that the first non-excluded maintainer entry is the one used
> for the automatic process.
This could even make the need for "contact=0|1" unneccessary (since at least 
one bugzilla account should be a valid assignee), yet it's of course better 
to still have it anyway.

> In light of the above, how about 'automatic=0'?
Kind of what I proposed, though I'd include "assign" and/or as  
Jan jkt Kundrát proposed "bug" somewhere in the variable name.
Like... "AutoBugAssign=whatever" or so.


Daniel
--
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Nguyen Thai Ngoc Duy]

On 4/26/07, Joshua Jackson <tsunam@gentoo.org> wrote:
> Nguyen Thai Ngoc Duy wrote:
> > On 4/26/07, Robin H. Johnson <robbat2@gentoo.org> wrote:
> >> Case 2 - Metadata contains a single maintainer
> >> ----------------------------------------------
> >> - The herd field is not used.
> >> - The maintainer address is used as the bugzilla assignee.
> >> This is important for all the herds that have aliases that are NOT the
> >> same as their herd name!
> >> This diverges from existing manual practice, to avoid unnecessary
> >> duplicate mail, and means that existing metadata may need a cleanup.
> >
> > It should take devaway into account.
> >
> why? Seriously, dev-away != dev retired... having it take devaway into
> account is pointless in my opinion as it won't improve it being properly
> assigned...as it'll be covered in other cases, and its not like there's
> not bugs for all of us dev's that have not sat there for a month or so,
> at some point

I meant if a maintainer is away, his/her herd should be assignee with
him/her CCed.

-- 
Duy
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Jakub Moc]

On 4/27/07, Nguyen Thai Ngoc Duy <pclouds@gmail.com> wrote:
> On 4/26/07, Joshua Jackson <tsunam@gentoo.org> wrote:
> > Nguyen Thai Ngoc Duy wrote:
> > > It should take devaway into account.
> > >
> > why? Seriously, dev-away != dev retired... having it take devaway into
> > account is pointless in my opinion as it won't improve it being properly
> > assigned...as it'll be covered in other cases, and its not like there's
> > not bugs for all of us dev's that have not sat there for a month or so,
> > at some point
>
> I meant if a maintainer is away, his/her herd should be assignee with
> him/her CCed.

Eh; if the maintainer has been away for months (a.k.a MIA), then
yeah... otherwise, no reason for this if someone's away for a week.
Sorry to disappoint you and others here, but the scripts will lack
artificial intelligence and frankly I don't see what exactly are you
expecting from this whole thing. Anyway, good luck. ;)

--
Jakub Moc
Email: jakub.moc@gmail.com
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Ned Ludd]

On Thu, 2007-04-26 at 22:01 -0700, Robin H. Johnson wrote:
> On Fri, Apr 27, 2007 at 02:33:50AM +0200, Danny van Dyk wrote:
> > > Both 'assign' and 'cc' (and derivations thereof are not suitable).
> > notification=assignment|cc|none ?
> This is to answer expose's question as well, but the attribute should
> only indicate if the maintainer entry should be used for any automatic
> process at all, not how to use it.
> 
> One of the reasons is that multiple maintainers each with
> notification=assignment obviously won't work, and it's non-trivial to
> validate via the DTD (yes, DTDs suck compared to XSchema, I know).
> 
> I intend that the first non-excluded maintainer entry is the one used
> for the automatic process.
> 
> In terms of implementing this in the DTD, I'm going to specify that
> 'contact=1' (or whatever name we settle on) is the default, so that we
> don't break validation of any existing metadata:
> 
> <!ATTLIST maintainer
>   contact   (0|1)   1   -- should this maintainer be used by 
>                         -- automatic processes?
>  >
> 
> In light of the above, how about 'automatic=0'?

Please keep with your original idea of letting maintainers opt out vs
some of the ideas proposed in this thread where maintainers have to opt
in as I'm sure the metadata.xml files wont be updated by enough people
to really gain the benefit of what we are trying to do here if they have
to do opt in.

Thanks.

 
-- 
Ned Ludd <solar@gentoo.org>

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Robin H. Johnson]

[-- Attachment #1: Type: text/plain, Size: 647 bytes --]

On Fri, Apr 27, 2007 at 10:32:25AM +0200, Jan Kundr?t wrote:
> AFAIK the preferred way of specifying boolean values in XML is to use
> contact="contact", not contact="1".
I can't find this described anywhere in the XML specification
http://www.w3.org/TR/REC-xml/

Have you got a reference for it?

> Speaking of name, I'd try to include string "bug" somewhere in the
> attribute name, like "bugzilla-auto-assignment" or something...
I'll follow this up in solar's post.

-- 
Robin Hugh Johnson
Gentoo Linux Developer & Council Member
E-Mail     : robbat2@gentoo.org
GnuPG FP   : 11AC BA4F 4778 E3F6 E4ED  F38E B27B 944E 3488 4E85

[-- Attachment #2: Type: application/pgp-signature, Size: 321 bytes --]

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Robin H. Johnson]

[-- Attachment #1: Type: text/plain, Size: 1317 bytes --]

On Fri, Apr 27, 2007 at 02:01:13PM +0200, expose@luftgetrock.net wrote:
> > I intend that the first non-excluded maintainer entry is the one used
> > for the automatic process.
> This could even make the need for "contact=0|1" unneccessary (since at least 
> one bugzilla account should be a valid assignee), yet it's of course better 
> to still have it anyway.
No, it doesn't make it unnecessary, as otherwise you have no way to
exclude a maintainer element.

1. Read maintainer blocks in order
2. Remove maintainer blocks that have 'contact=0'
3. First maintainer block is used for the assignee.
4. Remaining blocks go into the CC.

> > In light of the above, how about 'automatic=0'?
> Kind of what I proposed, though I'd include "assign" and/or as  
> Jan jkt Kundr??t proposed "bug" somewhere in the variable name.
> Like... "AutoBugAssign=whatever" or so.
No, because assign and CC are already overloaded terms.
If you approach it from a semantic angle, what does 'automatic-assign=0'
do? Does it mean that the maintainer is still used for the CC list?
That's why 'assign' and 'cc' should not occur in the attribute name.

-- 
Robin Hugh Johnson
Gentoo Linux Developer & Council Member
E-Mail     : robbat2@gentoo.org
GnuPG FP   : 11AC BA4F 4778 E3F6 E4ED  F38E B27B 944E 3488 4E85

[-- Attachment #2: Type: application/pgp-signature, Size: 321 bytes --]

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Robin H. Johnson]

[-- Attachment #1: Type: text/plain, Size: 916 bytes --]

On Fri, Apr 27, 2007 at 08:57:27AM -0700, Ned Ludd wrote:
> > In light of the above, how about 'automatic=0'?
> Please keep with your original idea of letting maintainers opt out vs
> some of the ideas proposed in this thread where maintainers have to opt
> in as I'm sure the metadata.xml files wont be updated by enough people
> to really gain the benefit of what we are trying to do here if they have
> to do opt in.
Err, nowhere in here have I said it was going to be opt-in.

Taking into account the other reasonable input, how about the name of
attribute 'automatic-bug' ?

'automatic-bug=1' will be implied by the DTD, and developers will have
to explicitly opt-out by including 'automatic-bug=0' in their
<maintainer> entries.

-- 
Robin Hugh Johnson
Gentoo Linux Developer & Council Member
E-Mail     : robbat2@gentoo.org
GnuPG FP   : 11AC BA4F 4778 E3F6 E4ED  F38E B27B 944E 3488 4E85

[-- Attachment #2: Type: application/pgp-signature, Size: 321 bytes --]

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Matti Bickel]

[-- Attachment #1: Type: text/plain, Size: 305 bytes --]

Robin H. Johnson <robbat2@gentoo.org> wrote:
> Taking into account the other reasonable input, how about the name of
> attribute 'automatic-bug' ?

I would like "assign" somewhere in the name, but i'd be fine with your
proposal as well.
-- 
Regards, Matti Bickel
Encrypted/Signed Email preferred

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Ned Ludd]

On Fri, 2007-04-27 at 10:57 -0700, Robin H. Johnson wrote:
> On Fri, Apr 27, 2007 at 08:57:27AM -0700, Ned Ludd wrote:
> > > In light of the above, how about 'automatic=0'?
> > Please keep with your original idea of letting maintainers opt out vs
> > some of the ideas proposed in this thread where maintainers have to opt
> > in as I'm sure the metadata.xml files wont be updated by enough people
> > to really gain the benefit of what we are trying to do here if they have
> > to do opt in.
> Err, nowhere in here have I said it was going to be opt-in.

With some of the dal and tri-state suggestions we have seen in this
thread automatic=1/contact=1 this would seem to be an opt-in vs opt-out.
But either way I don't care as long as we can get the bulk of the
bug-wranglers@ assigned to somewhere other than bug-wranglers@ and 
the scripts that are going to handle it don't have to become totally
convoluted while reasonable to maintain.


> Taking into account the other reasonable input, how about the name of
> attribute 'automatic-bug' ?

I don't see anything wrong with how it was proposed originally using
contact=0 


> 'automatic-bug=1' will be implied by the DTD, and developers will have
> to explicitly opt-out by including 'automatic-bug=0' in their
> <maintainer> entries.
> 
-- 
Ned Ludd <solar@gentoo.org>
Gentoo Linux

-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Jan Kundrát]

[-- Attachment #1: Type: text/plain, Size: 255 bytes --]

Robin H. Johnson wrote:
> Have you got a reference for it?

That's how it is in XHTML, so I thought it's common practice in XML as
well. That probably isn't true, so sorry for noise.

Cheers,
-jkt

-- 
cd /local/pub && more beer > /dev/mouth


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 252 bytes --]

Re: [gentoo-dev] Planning for automatic assignment of bugs

[expose]

Robin H. Johnson wrote:
> Taking into account the other reasonable input, how about the name of
> attribute 'automatic-bug' ?
Well, to complicate things even further, if you approach that from a semantic 
angle, automatic-bug is just as wrong as  the others, since no bug is 
automatically created...
Yet I am fine with (almost) anything that gives the user the idea of what it 
is all about: Bugzilla.
Therefore terms like bug, assignment, and maybe something like "automatic" are 
good choices.

So, although there's the cc-problem, I so agree with what Matti Bickel wrote
> I would like "assign" somewhere in the name, but i'd be fine with your
> proposal as well.
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[expose]

Ned Ludd wrote:
> I don't see anything wrong with how it was proposed originally using
> contact=0

The reason why "contact" isn't perfect was given by Mart leio Raudsepp yet, 
namely:
> contact=0 in metadata.xml in this context means that the automatic
> reassigning should not assign to that maintainer, but when a user looks
> whom to ask specific questions from and sees contact=0 he/she will
> understand he/she is not to contact that person as the value is zero,
> but Daniel wants them to contact precisely him in that case.
> A different keyword might be better for that reason.
-- 
gentoo-dev@gentoo.org mailing list

Re: [gentoo-dev] Planning for automatic assignment of bugs

[Flammie Pirinen]

[-- Attachment #1: Type: text/plain, Size: 789 bytes --]

2007-04-27, Robin H. Johnson sanoi:

> On Fri, Apr 27, 2007 at 10:32:25AM +0200, Jan Kundr?t wrote:
> > AFAIK the preferred way of specifying boolean values in XML is to
> > use contact="contact", not contact="1".
> I can't find this described anywhere in the XML specification
> http://www.w3.org/TR/REC-xml/
> 
> Have you got a reference for it?

That tradition stems from SGML, and in SGML it was also possible to
minimize this kind of true values by telling attribute without value.
The habit has carried over to XML even though it doesn't support
attribute minimization. The reference can be found in the SGML handbook
if you wish, I suppose.
-- 
Flammie, Gentoo Linux Documentation’s Finnish head translator
and FlameEyes’ bot <http://dev.gentoo.org/~flammie>.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 189 bytes --]