suggestion for jira plugin
First off, nice job. Finally able to see all incoming and outgoing
messages. Further nice that I can reply from NNTP client without having
to contort myself to get passed the jira interface.
Would like to suggest that:
1- You don't show null for submitter. I am assuming it's null because
of some import, or something happened where the data is not there. If
null show name of last person to update. It looks weird that way. I
noticed that something was resolved by null, when it looked obvious that
Maxim Mossienko had updated it.
2- I think it would be easier to read in the list if we had something
like: blabla, blablabla. If it's comment or resolved,
then blablabla, blablabla
Thanks for the hard work.
R
Please sign in to leave a comment.
Hum also just tested this... Should we be allowing the posting of new
tickets to jira via NNTP? It seems to work, but am not sure if this is
a good thing.
R
> 2- I think it would be easier to read in the list if we had something
> like: blabla, blablabla. If it's comment or resolved,
> then blablabla, blablabla
It better be bla-bla-bla, so that everything conerning one
issue will be in one thread with nice title.
In article <nomail-82A744.11333511032005@mail.intellij.net>,
Robert Sfeir <nomail@foobar.com> wrote:
Seems if I post an issue via NNTP Client to Jira, the issue is nowhere
to be found in jira. Mark, I'm not sure where this puppy ends up, but
you should adjust your code to not allow it unless someone is replying
to an existing issue.
R
On Fri, 11 Mar 2005 11:21:45 -0500, Robert Sfeir wrote:
Hmm, I've not seen that one, and havn't seen that in my usenet reader yet
- got a pointer to the message(s) at all? ( I'll go hunt anyway ). It
should be pulling out the fullname of the JIRA user.
So mention the issue-type field in the subject? I have a few changes that
Sergey's wanting which I'll try to get to over the weekend, so will look
at this as well if everyones keen?
On Fri, 11 Mar 2005 19:42:43 +0300, Mike Aizatsky (JetBrains) wrote:
That -should- occur anyway as I'm setting the References: header, I saw
that under my local tests against my own nntp server, but Pan doesn't seem
to be threading against JBs server ( although Thunderbird threaded them
fine.... ) odd.
On Fri, 11 Mar 2005 11:33:35 -0500, Robert Sfeir wrote:
Dependant on the configuration of JBs Jira. I'd vote no on issue creation
via NNTP ( no way to set version affects, components, etc. etc. etc. )
On Fri, 11 Mar 2005 12:04:18 -0500, Robert Sfeir wrote:
How did you know an issue was created?
On Sat, 12 Mar 2005 09:28:22 +1300, Mark Derricutt wrote:
Not hard to miss thou - yesterday those posts were coming out with "Please
don't reply", and now null. Something must have changed in the config
when Sergey fixed the repeating post issue.
I use the "Please don't reply" ( full-name configured in the service )
when theres no comment-author specified in the JIRA IssueEvent. However
there should be a reference in there for the user anyway. I'll jump back
into debug mode and start wandering the object structure.
On Sat, 12 Mar 2005 09:40:18 +1300, Mark Derricutt wrote:
Well duh - issueEvent.getRemoteUser().getFullName() :) One of those
stupid overlooked things, just because theres no comment (with associated
author) doesn't mean the datas not there :)
On Fri, 11 Mar 2005 11:21:45 -0500, Robert Sfeir wrote:
Robert,
Another idea(tm) I just had was this, much like we already have in the old
tracker groups, create subgroups of jetbrains.intellij.jira.idea with
(lowercased spaces turned into dots?) the issue type.
Then instead of the plugin posting to the top level newsgroup, there
broken down ( and maybe if the subgroup doesn't exist, post to the top
level ):
jetbrains.intellij.jira.newfeature
jetbrains.intellij.jira.bug
jetbrains.intellij.jira.performance.problem
jetbrains.intellij.jira.usability.problem
( The last two to me would be best dropping the word problem I guess, in
this situation anyway ).
Anyone else got any thoughts?
Don't complicate it. It's fine how it is. I guess it doesn't matter if
it's a feature or a bug.
In article <pan.2005.03.11.20.31.58.846674@gmail.com>,
Mark Derricutt <talios@gmail.com> wrote:
As I said, I don't believe the issue was created, but I can see it in
the NNTP client... so somehow it's there somewhere.
R
In article <pan.2005.03.11.20.31.01.836348@gmail.com>,
Mark Derricutt <talios@gmail.com> wrote:
Then it should be configured to not do so :) Cuz right now you sort of
can.
Mark Derricutt wrote:
I think it's fine the way it is. Having multiple newsgroups would be
too fine-grained given the larger number of issue types we have now.
By the way, Alain's issues have subjects such as Créée and Commenté
rather than Created and Commented. Is this intentional?
Mark Derricutt wrote:
In Mozilla 1.7.5 it seems to work fine as long as the "Created" message
has actually been sent out. When people update or comment older bugs
for which the "created" message hasn't been sent out, the References:
header seems to refer to a non-existing message ID, and then we just get
grouping by subject.
This could be solved by sending out "Created" messages for all bugs
currently in the database, but that might produce thousands of messages
all at once.
A possible workaround might be to check for each message whether a
"Created:" message for this bug has actually been sent out. If it has,
add a References: header referring to the "Created" message. If not,
give this message the ID that would normally have been given to the
"Created:" message, and omit the References: header.
On Sat, 12 Mar 2005 11:55:29 +0100, Jonas Kvarnström wrote:
Yep - agreed.
Not intentional no. I noticed that before and was sitting there going
"huh?". I'll have to ask Mike about that...
On Sat, 12 Mar 2005 01:26:50 -0500, Robert Sfeir wrote:
Ahh - that means its posted to the usenet server. Only messages that
mention a JIRA ticket get commented back into jira.
On Sat, 12 Mar 2005 12:06:44 +0100, Jonas Kvarnström wrote:
Ah yes - I just set the References to what I'd normally set for the
created previous message. That'll be it :)
or just not bother and see effectively the same result
Mark Derricutt wrote:
>>give this message the ID that would normally have been given to the
>>"Created:" message, and omit the References: header.
Not bother doing what? Giving it a References: header at all? I don't
think that will work, at least not if you intend to rely on pure
subject-based threading: Given that you add "Created:", "Commented:"
and so on to the subject, issues related to the same bug don't all share
the same subject.
In article <pan.2005.03.12.11.25.44.971276@gmail.com>,
Mark Derricutt <talios@gmail.com> wrote:
I'm not sure that's so good, the usefulness of this is that if someone
is commenting by NNTP that those comments be added to the ticket. If
someone changes the subject somehow, those comments are lost. I would
not rely on the subject, I would rely on information in the header.
R
In article <d0uhqu$nav$1@is.intellij.net>,
Jonas Kvarnström <jonkv@ida.liu.se> wrote:
This is because Alain is looking at it with a French browser, so Jira
deals with the il8n info that way, it doesn't seem to translate the
subject back to english when posting.
R
On Sat, 12 Mar 2005 13:35:18 +0100, Jonas Kvarnström wrote:
I meant not bother NOT adding it and tracking if I've already posted the
create to the newsgroup ( there's also no guarantee the create message
will be available in the server either, depending on how often messages
expire, if ever )
On Sat, 12 Mar 2005 10:17:51 -0500, Robert Sfeir wrote:
Unfortunately this is crossing over to Atlassians code, not mine. And the
message handler classes ( there behaviour is documented here:
http://www.atlassian.com/software/jira/docs/v3.1/services.html ).
Mark Derricutt wrote:
>>> or just not bother and see effectively the same result
>>Not bother doing what? Giving it a References: header at all?
I may be stupid but I still don't understand how you mean.
The problem that I see is that if the "Created" message for a particular
bug hasn't been sent, messages for that bug show up as one distinct
thread for each distinct subject, possibly quite far away from each
other in the thread list of the news reader: "Assigned: IDEA-827",
"Commented: IDEA-827", "Resolved: IDEA-827", "Reopened: IDEA-827".
There's no way of following everything that's related to a particular
bug. Because the issue number is not the first thing in the subject
line, I can't even sort the messages alphabetically and get any kind of
coherent view of these bugs. That still wouldn't work for bugs that
change IDs, such as IDEA-832 -> IDEADEV-578 (and I was quite impressed
that your news gateway does gather all messages about such bugs in the
same thread even after they are renamed/renumbered, as long as the
initial "Created:" message is presen).
In my suggestion, the first message actually posted about a bug would
not have a reference to the message ID of the "created" message (because
that message had never been sent), but would instead have that message
ID itself. Then, without any further changes, all other messages sent
about that bug would be linked under that first message, resolving the
problem that some bugs now
I don't see how "not bother" would give effectively the same result.
> ( there's also no guarantee the create message will be available
> in the server either, depending on how often messages expire, if
> ever )
There are 81376 messages in j.i.t.i.bugs, dating back to August 2002. I
don't think messages expire at all.
On Sat, 12 Mar 2005 22:38:37 +0100, Jonas Kvarnström wrote:
Not stupid no...
Correct...
It's not so much the presence of the "Created" message, but the presence
of its Message-ID in your newsclient.
I'll see what I can do with the JIRA MessageThread manager which tracks
Message-IDs, unfortunately, the MailThreadManager only exposes methods to
a) create a new thread, b) locate an issue from a message. It doesn't
expose any methods to retreive Message-ID's recorded against an issue (
which would make support this realllllly easy ).
I'll see if I get around that somehow.
That being said, theres not 81376 messages downloaded to my news client,
which would produce similar hanging / unthreaded posts if I hadn't
downloaded the parent message, or had my newsclient to show only unread
posts, not unread+thread posts.
As more issues are created in JIRA with the gateway active, this problem
won't really be an major issue, however I do agree it should be addressed
( I had briefly investigated it when originally writing the code, but at
the time couldn't see a way to get the recorded Message-ID's out ( without
hitting the database directly, which I'd rather not do if I can help it).
Mike Aizatsky (JetBrains) wrote:
I change the code to always use the english version the subject, but I
like this idea better anyway, so will change that and pass a new build
over to JB.
On Mon, 14 Mar 2005 11:08:42 +1300, Mark Derricutt wrote:
Doh - forgot to change the subject over. Anyway, forwarded a new build
over that looksup the first message sent by the issueposter and uses that
message-id for the original message id ( defaulting to the existing
assumed message id ).