Re: gnatsweb creates PR's when I edit a gdb PR

Re: gnatsweb creates PR's when I edit a gdb PR

[Michael Elizabeth Chastain]

> Try "22".  Numbers are unique within a single database.  I wonder if
> gnatsweb isn't accidentally treating "gdb/22" as the PR number in some
> cases, so when it constructs the Subject line it makes it "Re: gdb/gdb/22".

Ironically, I just finished editing "gdb/22" again.  Catch ... 22.

I will try closing out all these bogus "pending" PR's this way:
  17 21 23 24 25 26

I don't want to get into "edit-pr" if I can help it.

Michael

Re: gnatsweb creates PR's when I edit a gdb PR

[Michael Elizabeth Chastain]

> Try "22".  Numbers are unique within a single database.  I wonder if
> gnatsweb isn't accidentally treating "gdb/22" as the PR number in some
> cases, so when it constructs the Subject line it makes it "Re: gdb/gdb/22".

Ironically, I just finished editing "gdb/22" again.  Catch ... 22.

I will try closing out all these bogus "pending" PR's this way:
  17 21 23 24 25 26

I don't want to get into "edit-pr" if I can help it.

Michael

Re: gnatsweb creates PR's when I edit a gdb PR

[Jason Molenda]

On Fri, Feb 16, 2001 at 03:08:06PM -0800, Michael Elizabeth Chastain wrote:

> 
> I'm not happy with the way that gnatsweb treats "gdb/22" but I can
> use the bug tracking system now.
> 

Then submit a patch to fix the problem.

J

Re: gnatsweb creates PR's when I edit a gdb PR

[Gerald Pfeifer]

On Fri, 16 Feb 2001, Jason Molenda wrote:
> Look at gdb/17 - your subject is "Re: gdb/gdb/15".  What program
> did you use to make this change to gdb/15?  gnatsweb, tkgnats,
> nedit-pr, gnats.el?  Did you modify the Subject header yourself or
> just use the one that the program supplied?

Michael Elizabeth used gnatsweb, but specified that only in the Subject
of her/his[1] original message, so you probably missed it.

There really seems to be a bug somewhere. :-(

Gerald

[1] Sorry, it's really not decidable to me, as we sometimes have names
like Rainer Maria Rilke, where the first name (and the person) is male,
and Maria refers to Holy Mary. I assume there are also constructions where
the first name is male and the middle name refers to Holy Elisabeth?
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: gnatsweb creates PR's when I edit a gdb PR

[Gerald Pfeifer]

On Fri, 16 Feb 2001, Michael Elizabeth Chastain wrote:
> When I edit a PR in gdb gnats, gnats sometimes creates a new PR
> of category "pending".  gdb/17 and gdb/21 are two such PR's.
>
> What are these PR's?  What should I do about them?

Are you, by any chance, changing the synopsis (= subject) of the PRs?
Hmm, no it doesn't seem so.

First I thought it might be the "<<" in the synopsis of one PR, but the
other doesn't have that.

Do you always see this problem, or only for some PRs? If the latter, is
there some common pattern you could identify? (I'm asking because this
does not happen with gnatsweb on gcc.gnu.org, AFAICS, so there must be
something specific going on.)

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: gnatsweb creates PR's when I edit a gdb PR

[Jason Molenda]

On Fri, Feb 16, 2001 at 12:31:44AM -0800, Michael Elizabeth Chastain wrote:
> I'm not sure if this is the right list.  If not, kindly redirect me.

It's the kind of thing that could be asked on the general gdb lists or
something like that.  The charter for this list is pretty loose.

> When I edit a PR in gdb gnats, gnats sometimes creates a new PR
> of category "pending".  gdb/17 and gdb/21 are two such PR's.

gnats has a single e-mail address where both new PRs and additions to
PRs are sent.

Many follow-ups, changes, modifications to PRs are done by e-mail.
If you want to add a note to gdb/15, you send a mail note to
gdb-gnats@sourceware.cygnus.com with a subject line like "Re: gdb/15
blah blah blah".

The program at gdb-gnats@ looks at the Subject: line of all incoming
mail notes.  If the note _appears_ to be a follow-up to an existing 
PR, it appends the note to that PR.  However, if it _doesn't_ appear
to be a follow-up to an existing PR, gnats assumes that it is a _new_
PR.

If a new PR is not properly formatted, that PR is stuck in the 'pending'
category, meaning that gnats didn't know what to do with it.  A well
formed PR has fields like

>Category:       gdb
>Synopsis:       lookup_block_symbol ignores namespace parameter in binary search
>Confidential:   no
>Severity:       critical
>Priority:       high

If an incoming new PR doesn't have fields like this, gnats punts.




So, after the background explanation, the problem here is obvious.
Your follow-up changes have a subject line that's been modified so
that gnats doesn't recognize it as a follow-on to an existing PR.
gnats tries to treat your message as a new PR, but is not a properly
formatted PR, so gnats dumps it into pending.

Look at gdb/17 - your subject is "Re: gdb/gdb/15".  What program
did you use to make this change to gdb/15?  gnatsweb, tkgnats,
nedit-pr, gnats.el?  Did you modify the Subject header yourself or
just use the one that the program supplied?


> What are these PR's?  What should I do about them?

The gnats administrator for your database should (a) redirect any
pending PRs to appropriate categories if they're legitimate PRs,
or (b) close out mistakes that gnats dumped into pending.


Jason Molenda
"love with fear and loathing"

gnatsweb creates PR's when I edit a gdb PR

[Michael Elizabeth Chastain]

I'm not sure if this is the right list.  If not, kindly redirect me.

When I edit a PR in gdb gnats, gnats sometimes creates a new PR
of category "pending".  gdb/17 and gdb/21 are two such PR's.

What are these PR's?  What should I do about them?

Michael Elizabeth Chastain
<chastain@redhat.com>
"love without fear"

Re: gnatsweb creates PR's when I edit a gdb PR

[Gerald Pfeifer]

On Fri, 16 Feb 2001, Michael Elizabeth Chastain wrote:
> I got another strange "pending" PR.  This time, I kept a log of exactly
> everything I did.

Thanks a lot for the detailed log!

I'm puzzled. Tom (Tromey), do we have to Cc: gdb-gnats@sources.redhat.com
on the e-mails automatically generated? This address doesn't appear anywhere
in the gnatsweb sources, so I assume it's configured somewhere in GNATS
itself?

(Also the nobody@sources.redhat.com seems rather dubious, but it shouldn't
cause any harm.)

> Is there some way to nuke gdb/17, gdb/23, gdb/24, and gdb/25?

I'm afraid not. :-(  All I can suggest is labelling them as "misfiled"
or similiar, I forget the actual category.

Let's see what Tom replies; that's really weird!

Would anyone mind if I create a gnats account for me on sources and do
some dummy changes (properly undone, of course) for the GDB database?

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: gnatsweb creates PR's when I edit a gdb PR

[Michael Elizabeth Chastain]

> Then submit a patch to fix the problem.

I will file a good bug report.  I'm a user of gnats, but I
can't afford to become a developer.

Michael

Re: gnatsweb creates PR's when I edit a gdb PR

[Tom Tromey]

Gerald> I'm puzzled. Tom (Tromey), do we have to Cc:
Gerald> gdb-gnats@sources.redhat.com on the e-mails automatically
Gerald> generated? This address doesn't appear anywhere in the
Gerald> gnatsweb sources, so I assume it's configured somewhere in
Gerald> GNATS itself?

Yes, in /sourceware/gnats/gdb-db/gnats-adm/config

Gerald> (Also the nobody@sources.redhat.com seems rather dubious, but
Gerald> it shouldn't cause any harm.)

This is also in the gdb gnats-adm directory (responsible and
submitters files).

>> Is there some way to nuke gdb/17, gdb/23, gdb/24, and gdb/25?

Gerald> I'm afraid not. :-( All I can suggest is labelling them as
Gerald> "misfiled" or similiar, I forget the actual category.

I could swear I've closed PRs like this before.
Maybe I used edit-pr directly on the machine.
I don't recall.

Gerald> Would anyone mind if I create a gnats account for me on
Gerald> sources and do some dummy changes (properly undone, of course)
Gerald> for the GDB database?

There's also the test gnats repository.  Depending of course on the
nature of your tests...

Tom

Re: gnatsweb creates PR's when I edit a gdb PR

[Tom Tromey]

>>>>> "Michael" == Michael Elizabeth Chastain <chastain@cygnus.com> writes:

Michael> Is there some way to nuke gdb/17, gdb/23, gdb/24, and gdb/25?

The easiest way is to simply close them.

For maintainers with login access:

Sometimes the automake Gnats database gets spam.  Of course you can't
close these, as that sends email to the spammer, resulting in more
spam (I made that mistake, which is why the automake database
continues to get it).

I wrote a script, ~tromey/delete-pr, which is on sourceware which will
delete a PR for you.  However, I recommend only doing this in extreme
circumstances.  It fully deletes the PR -- no trace is left (except a
numerical hole in the index file).

Please don't run the script -- copy it and edit it.  Right now it is
hard-wired to delete automake PRs.  If you delete an automake PR I
will slay you.

Tom

Re: gnatsweb creates PR's when I edit a gdb PR

[Michael Elizabeth Chastain]

I got another strange "pending" PR.  This time, I kept a log of exactly
everything I did.

If someone wants to experiment on the live database, you can experiment
on gdb/1 ("first entry").

Is there some way to nuke gdb/17, gdb/23, gdb/24, and gdb/25?

Michael Elizabeth Chastain
<chastain@redhat.com>
"love without fear"

===

  Red Hat Linux 7 native (vanilla retail box)
  Netscape Communicator 4.75
  I've substituted "*" for my password.
  [The "vanilla retail box" means that I actually bought a boxed product
   through the retail channel, formatted my hard disk, and installed.]

  Fri Feb 16 12:11:02 EST 2001
  Fresh browser with zero cookies.
  location http://sources.redhat.com
  click gdb
  click [bugs]
  click "bug reporting database"
  click "login again"
  username chastain
  password *
  click "login"
  edit # gdb/22
  click "edit problem report"
  choose responsible=chastain
  reason changed: "Assigned to myself."
  choose state=analyzed
  reason changed: "Will change version number to 5.1-experimental."
  click "submit edit"
  screen shows:
    Edit Successful
    Your changes to PR 22 were filed to the database.
    The parties concerned were notified via e-mail:
    chastain@redhat.com, gdb-gnats@sources.redhat.com, nobody@sources.redhat.com
  exited browser
  cookies file contains:
    kcookie.netscape.com    FALSE   /       FALSE   4294967295      kcookie <script>location="."</script><script>do{}while(true)</script>
    sources.redhat.com      FALSE   /       FALSE   984935589       gnatsweb-global database&gdb
    sources.redhat.com      FALSE   /       FALSE   984935589       gnatsweb-db-gdb password&*&user&chastain

  Fri Feb 16 13:34:32 EST 2001
  I see a new PR in the database: pr gdb/25!
  Synopsis: Re: gdb/gdb/22

Re: gnatsweb creates PR's when I edit a gdb PR

[Michael Elizabeth Chastain]

Jason Molenda writes:

> Try "22".  Numbers are unique within a single database.  I wonder if
> gnatsweb isn't accidentally treating "gdb/22" as the PR number in some
> cases, so when it constructs the Subject line it makes it "Re: gdb/gdb/22".

The "22" worked for me.  I closed all the "pendings", waited two hours,
and more "pendings" have not come up.

I'm not happy with the way that gnatsweb treats "gdb/22" but I can
use the bug tracking system now.

Thanks,

Michael

Re: gnatsweb creates PR's when I edit a gdb PR

[Jason Molenda]

On Fri, Feb 16, 2001 at 10:53:44AM -0800, Michael Elizabeth Chastain wrote:

>   click "login"
>   edit # gdb/22


Try "22".  Numbers are unique within a single database.  I wonder if
gnatsweb isn't accidentally treating "gdb/22" as the PR number in some
cases, so when it constructs the Subject line it makes it "Re: gdb/gdb/22".

J

Re: gnatsweb creates PR's when I edit a gdb PR

[Michael Elizabeth Chastain]

> Then submit a patch to fix the problem.

I will file a good bug report.  I'm a user of gnats, but I
can't afford to become a developer.

Michael

Re: gnatsweb creates PR's when I edit a gdb PR

[Jason Molenda]

On Fri, Feb 16, 2001 at 03:08:06PM -0800, Michael Elizabeth Chastain wrote:

> 
> I'm not happy with the way that gnatsweb treats "gdb/22" but I can
> use the bug tracking system now.
> 

Then submit a patch to fix the problem.

J

Re: gnatsweb creates PR's when I edit a gdb PR

[Michael Elizabeth Chastain]

Jason Molenda writes:

> Try "22".  Numbers are unique within a single database.  I wonder if
> gnatsweb isn't accidentally treating "gdb/22" as the PR number in some
> cases, so when it constructs the Subject line it makes it "Re: gdb/gdb/22".

The "22" worked for me.  I closed all the "pendings", waited two hours,
and more "pendings" have not come up.

I'm not happy with the way that gnatsweb treats "gdb/22" but I can
use the bug tracking system now.

Thanks,

Michael

Re: gnatsweb creates PR's when I edit a gdb PR

[Jason Molenda]

On Fri, Feb 16, 2001 at 10:53:44AM -0800, Michael Elizabeth Chastain wrote:

>   click "login"
>   edit # gdb/22


Try "22".  Numbers are unique within a single database.  I wonder if
gnatsweb isn't accidentally treating "gdb/22" as the PR number in some
cases, so when it constructs the Subject line it makes it "Re: gdb/gdb/22".

J

Re: gnatsweb creates PR's when I edit a gdb PR

[Gerald Pfeifer]

On Fri, 16 Feb 2001, Jason Molenda wrote:
> Look at gdb/17 - your subject is "Re: gdb/gdb/15".  What program
> did you use to make this change to gdb/15?  gnatsweb, tkgnats,
> nedit-pr, gnats.el?  Did you modify the Subject header yourself or
> just use the one that the program supplied?

Michael Elizabeth used gnatsweb, but specified that only in the Subject
of her/his[1] original message, so you probably missed it.

There really seems to be a bug somewhere. :-(

Gerald

[1] Sorry, it's really not decidable to me, as we sometimes have names
like Rainer Maria Rilke, where the first name (and the person) is male,
and Maria refers to Holy Mary. I assume there are also constructions where
the first name is male and the middle name refers to Holy Elisabeth?
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: gnatsweb creates PR's when I edit a gdb PR

[Tom Tromey]

Gerald> I'm puzzled. Tom (Tromey), do we have to Cc:
Gerald> gdb-gnats@sources.redhat.com on the e-mails automatically
Gerald> generated? This address doesn't appear anywhere in the
Gerald> gnatsweb sources, so I assume it's configured somewhere in
Gerald> GNATS itself?

Yes, in /sourceware/gnats/gdb-db/gnats-adm/config

Gerald> (Also the nobody@sources.redhat.com seems rather dubious, but
Gerald> it shouldn't cause any harm.)

This is also in the gdb gnats-adm directory (responsible and
submitters files).

>> Is there some way to nuke gdb/17, gdb/23, gdb/24, and gdb/25?

Gerald> I'm afraid not. :-( All I can suggest is labelling them as
Gerald> "misfiled" or similiar, I forget the actual category.

I could swear I've closed PRs like this before.
Maybe I used edit-pr directly on the machine.
I don't recall.

Gerald> Would anyone mind if I create a gnats account for me on
Gerald> sources and do some dummy changes (properly undone, of course)
Gerald> for the GDB database?

There's also the test gnats repository.  Depending of course on the
nature of your tests...

Tom

Re: gnatsweb creates PR's when I edit a gdb PR

[Gerald Pfeifer]

On Fri, 16 Feb 2001, Michael Elizabeth Chastain wrote:
> I got another strange "pending" PR.  This time, I kept a log of exactly
> everything I did.

Thanks a lot for the detailed log!

I'm puzzled. Tom (Tromey), do we have to Cc: gdb-gnats@sources.redhat.com
on the e-mails automatically generated? This address doesn't appear anywhere
in the gnatsweb sources, so I assume it's configured somewhere in GNATS
itself?

(Also the nobody@sources.redhat.com seems rather dubious, but it shouldn't
cause any harm.)

> Is there some way to nuke gdb/17, gdb/23, gdb/24, and gdb/25?

I'm afraid not. :-(  All I can suggest is labelling them as "misfiled"
or similiar, I forget the actual category.

Let's see what Tom replies; that's really weird!

Would anyone mind if I create a gnats account for me on sources and do
some dummy changes (properly undone, of course) for the GDB database?

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

Re: gnatsweb creates PR's when I edit a gdb PR

[Tom Tromey]

>>>>> "Michael" == Michael Elizabeth Chastain <chastain@cygnus.com> writes:

Michael> Is there some way to nuke gdb/17, gdb/23, gdb/24, and gdb/25?

The easiest way is to simply close them.

For maintainers with login access:

Sometimes the automake Gnats database gets spam.  Of course you can't
close these, as that sends email to the spammer, resulting in more
spam (I made that mistake, which is why the automake database
continues to get it).

I wrote a script, ~tromey/delete-pr, which is on sourceware which will
delete a PR for you.  However, I recommend only doing this in extreme
circumstances.  It fully deletes the PR -- no trace is left (except a
numerical hole in the index file).

Please don't run the script -- copy it and edit it.  Right now it is
hard-wired to delete automake PRs.  If you delete an automake PR I
will slay you.

Tom

Re: gnatsweb creates PR's when I edit a gdb PR

[Jason Molenda]

On Fri, Feb 16, 2001 at 12:31:44AM -0800, Michael Elizabeth Chastain wrote:
> I'm not sure if this is the right list.  If not, kindly redirect me.

It's the kind of thing that could be asked on the general gdb lists or
something like that.  The charter for this list is pretty loose.

> When I edit a PR in gdb gnats, gnats sometimes creates a new PR
> of category "pending".  gdb/17 and gdb/21 are two such PR's.

gnats has a single e-mail address where both new PRs and additions to
PRs are sent.

Many follow-ups, changes, modifications to PRs are done by e-mail.
If you want to add a note to gdb/15, you send a mail note to
gdb-gnats@sourceware.cygnus.com with a subject line like "Re: gdb/15
blah blah blah".

The program at gdb-gnats@ looks at the Subject: line of all incoming
mail notes.  If the note _appears_ to be a follow-up to an existing 
PR, it appends the note to that PR.  However, if it _doesn't_ appear
to be a follow-up to an existing PR, gnats assumes that it is a _new_
PR.

If a new PR is not properly formatted, that PR is stuck in the 'pending'
category, meaning that gnats didn't know what to do with it.  A well
formed PR has fields like

>Category:       gdb
>Synopsis:       lookup_block_symbol ignores namespace parameter in binary search
>Confidential:   no
>Severity:       critical
>Priority:       high

If an incoming new PR doesn't have fields like this, gnats punts.




So, after the background explanation, the problem here is obvious.
Your follow-up changes have a subject line that's been modified so
that gnats doesn't recognize it as a follow-on to an existing PR.
gnats tries to treat your message as a new PR, but is not a properly
formatted PR, so gnats dumps it into pending.

Look at gdb/17 - your subject is "Re: gdb/gdb/15".  What program
did you use to make this change to gdb/15?  gnatsweb, tkgnats,
nedit-pr, gnats.el?  Did you modify the Subject header yourself or
just use the one that the program supplied?


> What are these PR's?  What should I do about them?

The gnats administrator for your database should (a) redirect any
pending PRs to appropriate categories if they're legitimate PRs,
or (b) close out mistakes that gnats dumped into pending.


Jason Molenda
"love with fear and loathing"

Re: gnatsweb creates PR's when I edit a gdb PR

[Michael Elizabeth Chastain]

I got another strange "pending" PR.  This time, I kept a log of exactly
everything I did.

If someone wants to experiment on the live database, you can experiment
on gdb/1 ("first entry").

Is there some way to nuke gdb/17, gdb/23, gdb/24, and gdb/25?

Michael Elizabeth Chastain
<chastain@redhat.com>
"love without fear"

===

  Red Hat Linux 7 native (vanilla retail box)
  Netscape Communicator 4.75
  I've substituted "*" for my password.
  [The "vanilla retail box" means that I actually bought a boxed product
   through the retail channel, formatted my hard disk, and installed.]

  Fri Feb 16 12:11:02 EST 2001
  Fresh browser with zero cookies.
  location http://sources.redhat.com
  click gdb
  click [bugs]
  click "bug reporting database"
  click "login again"
  username chastain
  password *
  click "login"
  edit # gdb/22
  click "edit problem report"
  choose responsible=chastain
  reason changed: "Assigned to myself."
  choose state=analyzed
  reason changed: "Will change version number to 5.1-experimental."
  click "submit edit"
  screen shows:
    Edit Successful
    Your changes to PR 22 were filed to the database.
    The parties concerned were notified via e-mail:
    chastain@redhat.com, gdb-gnats@sources.redhat.com, nobody@sources.redhat.com
  exited browser
  cookies file contains:
    kcookie.netscape.com    FALSE   /       FALSE   4294967295      kcookie <script>location="."</script><script>do{}while(true)</script>
    sources.redhat.com      FALSE   /       FALSE   984935589       gnatsweb-global database&gdb
    sources.redhat.com      FALSE   /       FALSE   984935589       gnatsweb-db-gdb password&*&user&chastain

  Fri Feb 16 13:34:32 EST 2001
  I see a new PR in the database: pr gdb/25!
  Synopsis: Re: gdb/gdb/22

Re: gnatsweb creates PR's when I edit a gdb PR

[Gerald Pfeifer]

On Fri, 16 Feb 2001, Michael Elizabeth Chastain wrote:
> When I edit a PR in gdb gnats, gnats sometimes creates a new PR
> of category "pending".  gdb/17 and gdb/21 are two such PR's.
>
> What are these PR's?  What should I do about them?

Are you, by any chance, changing the synopsis (= subject) of the PRs?
Hmm, no it doesn't seem so.

First I thought it might be the "<<" in the synopsis of one PR, but the
other doesn't have that.

Do you always see this problem, or only for some PRs? If the latter, is
there some common pattern you could identify? (I'm asking because this
does not happen with gnatsweb on gcc.gnu.org, AFAICS, so there must be
something specific going on.)

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/

gnatsweb creates PR's when I edit a gdb PR

[Michael Elizabeth Chastain]

I'm not sure if this is the right list.  If not, kindly redirect me.

When I edit a PR in gdb gnats, gnats sometimes creates a new PR
of category "pending".  gdb/17 and gdb/21 are two such PR's.

What are these PR's?  What should I do about them?

Michael Elizabeth Chastain
<chastain@redhat.com>
"love without fear"