Special meeting on December 17, 2008
Agenda
- Attendance
- Announce agenda
- Should we require discussion solicitation for all issues which will go into the trunk? (discussion and vote)
- What method should be used use to push feature discussions to the email list? (automatic JIRA email, manual emails, digest emails?)
- Summary and closing
Notes
Attendance
- tclarke
- mconsidi
- kstreith
- dsulgrov
- rforehan
- gwelch
- dadkins
- raevans
- goffena
Summary
Discussion continued from the previous meeting on the first vote. Multiple points were made until attendees seemed satisfied with the discussion and a vote was held. Based on the results of the first vote, the second vote question was obsolete as it was contingent on approval of the first question.
Decisions
Should we require discussion solicitation for all issues which will go into the trunk? The motion failed. (9 attending, 1 yes, 5 nea, 3 abstain)
Logs
| *** | Opened channel log for #opticks at 12/18/2008 2:03:24 PM | |
| 14:03 | tclarke | logging is enabled |
| 14:03 | *** | mconsidi (n=mconsidi@12.188.157.129) has joined #opticks |
| 14:03 | tclarke | lets take attendance |
| 14:03 | tclarke | here |
| 14:03 | mconsidi | here |
| 14:03 | kstreith | here |
| 14:03 | dsulgrov | here |
| 14:03 | rforehan | here |
| 14:03 | gwelch | Here |
| 14:03 | dadkins | here |
| 14:04 | raevans | here |
| 14:04 | tclarke | here's the agenda |
| 14:04 | tclarke | # Attendance |
| 14:04 | tclarke | # Announce agenda |
| 14:04 | tclarke | # Should we require discussion solicitation for all issues which will go into the trunk? (discussion and vote) |
| 14:04 | tclarke | # What method should be used use to push feature discussions to the email list? (automatic JIRA email, manual emails, digest emails?) |
| 14:04 | tclarke | # Summary and closing |
| 14:04 | gwelch | On Monday, we may not have heard from all would-be participants |
| 14:05 | goffena | here if not too late |
| 14:05 | tclarke | btw, gwelch will be moderating today |
| 14:05 | goffena | or maybe i'm not to participate |
| 14:05 | gwelch | of course you should, if you'd like |
| 14:06 | gwelch | Does anyone want/need a recap of Monday's discussion? |
| 14:06 | tclarke | as a review, the question we are dealing with |
| 14:06 | tclarke | is show we require all jira issues have an associated mailing list thread |
| 14:07 | tclarke | if not, should there be a rule (all new features are required, for example) |
| 14:07 | tclarke | or should the developer working the issue decide if a thread is needed |
| 14:07 | gwelch | Thanks, tclarke. |
| 14:07 | gwelch | Ground rules for today's discussion . . . |
| 14:08 | gwelch | If you are in the middle of a thought, end your line with ". . ." . . . |
| 14:08 | *** | tjohnson (i=0cbc9d81@gateway/web/ajax/mibbit.com/x-4d083a73ccd18d2e) has joined #opticks |
| 14:08 | gwelch | If you're done, end with a "." . . . |
| 14:08 | rforehan | Seems to me that if a bug fix is transparent to users, there's no need for thread, e.g., fixing crash from attempt to access a null pointer. |
| 14:08 | tjohnson | Hola! . |
| 14:08 | tclarke | which suggests that if you have a long thought.....get something written to the chat room so people know you are talking |
| 14:09 | gwelch | Hopefully we won't walk on each other too much that way . . . |
| 14:09 | tclarke | ok, who would... |
| 14:09 | tclarke | like to summarize the argument for all jira issues? |
| 14:10 | gwelch | Do you want to, tclarke? |
| 14:10 | tclarke | ok.... |
| 14:11 | tclarke | the main argument for all issues having a thread is that the developer working the issue may not... |
| 14:11 | tclarke | know all the impact of a change, even if it should not affect the API... |
| 14:11 | tclarke | i.e. a plug-in might be relying on undocumented behavior, etc. |
| 14:12 | tclarke | the main arguments against all issues... |
| 14:12 | tclarke | is a) it takes extra time to wait for discussion... |
| 14:12 | tclarke | and b) there's a lower signal to noise on the mailing list which could cause readers to skip over potentially important discussion |
| 14:12 | tclarke | do I generally have those correct? |
| 14:13 | tclarke | if so, let's discuss |
| 14:13 | kstreith | i would add a "for" argument... |
| 14:13 | kstreith | consistency |
| 14:13 | kstreith | if we only discuss some of the issues on the mailing list, how does a new entrant to the mailing list, know that |
| 14:13 | kstreith | depending on the volume they may assume we are talking about everything, when in fact we are not. |
| 14:14 | tclarke | I'll add some information... |
| 14:14 | tclarke | that may lessen the for argument... |
| 14:14 | tclarke | JIRA has a capability to generate RSS feeds of issues (searches) which act almost like a "push"... |
| 14:14 | tclarke | and you can also watch an issue to get emails whenever that issue changes... |
| 14:15 | tclarke | we could put this information on the web site, possibly even a portlet which shows the past X new issue transistions |
| 14:16 | dsulgrov | I am not in favor of posting information on all issues to the dev mailing list... |
| 14:16 | dsulgrov | here's why... |
| 14:16 | dsulgrov | I agree that information should be posted to the developers mailing list when it affects developers... |
| 14:17 | dsulgrov | there are JIRA issues, however that do not affect external plug-in developers... |
| 14:17 | dsulgrov | e.g. OPTICKS-294, involving minor code style changes |
| 14:17 | dsulgrov | ... |
| 14:18 | dsulgrov | I think that any information posted to the developer mailing list should be highly relevant to external developers... |
| 14:18 | dsulgrov | as such, we may be able to establish several guidelines as to when information should be posted... |
| 14:19 | dsulgrov | two such guidelines would be as follows... |
| 14:19 | dsulgrov | when we are faced with a design decision that could take one of several paths... |
| 14:19 | dsulgrov | and we are not necessarily sold on any one path... |
| 14:20 | dsulgrov | another guideline would be when we are modifying an interface such that plug-in developers would need to rework existing code... |
| 14:21 | dsulgrov | I think that streamlining this information will help to achieve a balance between posting useful information and not requiring too much extra work for the committers |
| 14:21 | dsulgrov | finished |
| 14:21 | gwelch | counter-points? |
| 14:22 | tjohnson | It sounds like you are advocating discussion for source or functionally breaking changes and not for others... |
| 14:22 | tjohnson | I can see the merit in this. |
| 14:22 | dsulgrov | yes |
| 14:22 | kstreith | ok, let me revisit a few points |
| 14:23 | kstreith | dsulgrov is brining up the point of relevancy |
| 14:23 | goffena | like what dsulgrov is thinking. |
| 14:23 | kstreith | which can potentially be intermingled with tclarke's argument of signal to noise ratio and also tclarke's dicussion of splitting the mailing list into two |
| 14:24 | kstreith | so, the question is should we not post this information to any mailing list |
| 14:24 | kstreith | or should we make two mailing lists? |
| 14:24 | kstreith | hold on.. |
| 14:24 | kstreith | i have another separate point... |
| 14:25 | kstreith | tclarke made the argument against being that developers have to "wait" for the discussion on the mailing list before they can proceed |
| 14:25 | kstreith | i think that needs to be discussed, but i do not believe that is strictly required in all cases... |
| 14:25 | kstreith | ok, done. |
| 14:26 | gwelch | kstrieth, when would it not be required? |
| 14:26 | kstreith | ok... |
| 14:26 | kstreith | if we have a 8 week development cycle, let's say... |
| 14:26 | kstreith | if you are working on a change in week one... |
| 14:26 | kstreith | you could post the mailing list about it... |
| 14:27 | kstreith | but go ahead and proceed and commit that to the trunk without waiting for a resolution on the mailing list... |
| 14:27 | kstreith | then if someone brought up a valid point in let's say week 6 of development... |
| 14:27 | kstreith | we could decide to put another bug in and fix that problem in week 6... |
| 14:27 | kstreith | and then commit that to trunk in week 6... |
| 14:28 | kstreith | we've still accomplished the goal which is to make the best possible release in a 8 week period... |
| 14:28 | kstreith | now, obviously a developer has to make a judgement call based upon the code they are working on... |
| 14:28 | kstreith | and how likely they think it is that someone will point out a change later... |
| 14:29 | kstreith | that they think they should make... |
| 14:29 | kstreith | and obviously for large changes, you would want to wait until the discussion is complete on the mailing list... |
| 14:29 | kstreith | but it is not strictly required to wait. |
| 14:29 | kstreith | done. |
| 14:29 | tclarke | it seems to me that not waiting would defeat the main point of this exercise which is to avoid having to redesign/reimplement after the fact based on community feedback |
| 14:29 | goffena | my thoughts on emailing about each jira issue... |
| 14:30 | kstreith | the point is to get the best source code in a 8 week period. |
| 14:30 | goffena | If you must post for each JIRA issue please limit to something similar to one of the following:... |
| 14:30 | goffena | 1) Send one daily email which contains links to all JIRA issues updated for the day... |
| 14:30 | goffena | 2) Send one daily email which contains links to all JIRA issues created for the day... |
| 14:30 | goffena | 3) Send one email each time a JIRA issue is created, with a link to the JIRA issue... |
| 14:30 | goffena | In either case state in the pushed email notification that developers may log onto IRC and chat about their input/concerns. |
| 14:30 | goffena | This would makes use of JIRA, IRC and emails in a reasonable fasion, in my opinion. |
| 14:30 | goffena | done. |
| 14:30 | tclarke | that's why I suggested rss... |
| 14:31 | tclarke | as it allows people to generate those sorts of lists... |
| 14:31 | tclarke | using whichever reader they desire (IE and Firefox have built-in support and there are lots of aggregators like bloglines and google reader) |
| 14:31 | dsulgrov | kstreith: did you have anything in mind for separate mailing lists? |
| 14:31 | kstreith | in the previous meeting |
| 14:32 | kstreith | nevermind what I was about to say... |
| 14:32 | kstreith | goffena: as an aside, i think you meant something besides when an "issue is created", i think you meant when an "issue is being worked". |
| 14:33 | tjohnson | What is the point of the discussion on the dev list? People who are really, really interested will watch JIRA issues, participate in IRC meetings, watch RSS feeds, etc. I think the dev list is to throw a wider net and get to people who generally are too busy to do all of that. As such, it needs to be focused so the SNR doesn't get too low. |
| 14:33 | * | dsulgrov agrees with tjohnson |
| 14:33 | goffena | 1) for when "issue is being worked" |
| 14:33 | goffena | 2 and 3) only for when "created" |
| 14:34 | goffena | ... |
| 14:35 | goffena | once the developer knows about the issue, they could use JIRA to send them emails about the updated issue (i think that is currently possible). That way only the concerned developers subscribe to the JIRA issue, and only the concerned developer continues to receive emails on updates |
| 14:35 | goffena | . |
| 14:35 | goffena | agreeing with tjohnson |
| 14:36 | dsulgrov | my thought about separate mailing lists is as follows... |
| 14:36 | dsulgrov | we have a developer mailing list that should contain information relevant to developers... |
| 14:36 | dsulgrov | we also have a users mailing list that should contain info relevant to users... |
| 14:37 | dsulgrov | I'm having a difficult time imagining what other list should be made available... |
| 14:37 | dsulgrov | because less interested parties can check the road map, watch JIRA issues, etc. |
| 14:37 | dsulgrov | done |
| 14:37 | tclarke | the other possibility would be a core developer's mailing list... |
| 14:38 | tclarke | with in depth discussion on stuff plug-in devs don't care about... |
| 14:38 | tclarke | this is generally done with multiple cc's at this point. |
| 14:38 | tclarke | . |
| 14:38 | dsulgrov | who would subscribe? |
| 14:38 | tclarke | anyone with commit privs |
| 14:38 | tclarke | . |
| 14:38 | kstreith | this is used when you have active core contributors that are no singlely located |
| 14:38 | tclarke | I'm not necessarily advocating such a list, but it would be the obvious addition |
| 14:38 | tclarke | . |
| 14:39 | kstreith | since currenly all of core commiters are located in the same area, we haven't had a need for such a list |
| 14:39 | dsulgrov | kstreith: sure, but IRC seems like a good way to accommodate that |
| 14:39 | kstreith | yes, i agree and that is what we have been using |
| 14:39 | kstreith | but one could also use a mailing list |
| 14:39 | kstreith | done. |
| 14:40 | tclarke | dsulgrov: I tend to agree at this point and think another list would be superfluous, but it's an option that I thought should be mentioned |
| 14:40 | tclarke | . |
| 14:40 | tclarke | it's also possible to have an IRC bot that sends an IRC message when certain JIRA transitions occur |
| 14:40 | tclarke | . |
| 14:40 | tclarke | they have one on the various osgeo channels for example |
| 14:40 | tclarke | . |
| 14:41 | kstreith | in the interest of time, am i the only one if the camp to send all messages? |
| 14:41 | kstreith | if someone else in the camp, with me, please speak up |
| 14:41 | rforehan | I have to leave shortly - I'm in favor of enabling RSS feeds from JIRA so people can monitor issues of interest, but against putting every issue on e-mail. |
| 14:41 | kstreith | otherwise, I'll be quiet and we can move on |
| 14:41 | tclarke | we can vote if people think they know where they stand |
| 14:42 | dsulgrov | do mconsidi, raevans, or dadkins have any thoughts? |
| 14:42 | kstreith | i'm not asking for a vote at the moment, tclarke |
| 14:42 | tclarke | the vote would be for or against all issues to the mailing list |
| 14:42 | tclarke | . |
| 14:42 | gwelch | other points from anyone? |
| 14:42 | kstreith | i still have a point i wish to make before a vote, i just want to open the floor to others that might be in the same camp as I |
| 14:42 | kstreith | . |
| 14:43 | rforehan | I have to leave now - mark me down as against all issues |
| 14:44 | gwelch | thanks rforehand |
| 14:44 | dadkins | Enabling RSS feeds on JIRA would allow interested parties to see activity on all issues, which is functionally equivalent to having all issues on the mailing list, correct? |
| 14:44 | kstreith | dadkins, not exactly, you can't have discussion inside JIRA |
| 14:44 | kstreith | it is a one-way communication tool |
| 14:44 | kstreith | done. |
| 14:45 | dadkins | but someone could start a thread if desired. |
| 14:45 | tclarke | perhaps we should think about re-enabling comments in JIRA if we decide we need two way communication on issues |
| 14:45 | tclarke | . |
| 14:45 | tclarke | a discussion for another time, but a possibility |
| 14:45 | tclarke | . |
| 14:45 | mconsidi | I'm waiting for kstreith to make his last point. |
| 14:46 | tclarke | as for RSS capability, you can generate a feed from any JIRA search |
| 14:46 | tclarke | updates to specific tasks currently require a "Watch" |
| 14:46 | tclarke | . |
| 14:46 | kstreith | ok, my last point in the discussion for "all" or "some"... |
| 14:47 | kstreith | i hope we can all agree that none of has any great experience in opening up a large scale development effort (i.e. 4.3.X) to collaboration with anyone in the world... |
| 14:47 | *** | gmartin_cn is n=gmartin-@nat1.sp.collab.net (Guy Martin) |
| 14:47 | *** | gmartin_cn is on channels: #opticks |
| 14:47 | *** | gmartin_cn _is on server irc.freenode.net (http://freenode.net/)_ |
| 14:47 | *** | gmartin_cn has been idle for 1 hour 16 minutes 45 seconds |
| 14:48 | kstreith | and so I'm afraid that we may be anticipating problems that we haven't yet proven are there... |
| 14:48 | kstreith | i.e. in code speak, we make think the iter++ should be changed to ++iter to improve performance, but we haven't yet put any timing statements in... |
| 14:48 | kstreith | so, if you're on the fence, please consider trying the "all" just for this next release... |
| 14:49 | kstreith | that way we can get actual data points, and then in the future we can decide whether "all" or "some" is the right way... |
| 14:49 | kstreith | but with actual data points... |
| 14:49 | kstreith | now, some will say that we can just go the "some" route and get the same data points... |
| 14:50 | kstreith | i would point out that going the "all" route is the more daring route and the one that requires the least amount of training on the part of new entrants to the community... |
| 14:50 | kstreith | ok, i'm done. |
| 14:50 | tclarke | the one counterpoint I make is that we do (sort of) have a data point... |
| 14:51 | tclarke | it's known that the human brain ignores background noise... |
| 14:51 | tclarke | and that a data stream with a low signal to noise ratio will be largely ignored due to this (think lots of spam email...that one important one can slip through easily)... |
| 14:52 | tclarke | the think we are not 100% sure on is if people on the mailing list will consider these extra message as a significant noise source to cause the data stream to be ignored |
| 14:52 | tclarke | . |
| 14:52 | raevans | I like having RSS feeds on JIRA, but I don't like requiring to put all issues up on the mailing list... |
| 14:52 | raevans | since there are some issues that we don't need to have any discussion on. |
| 14:53 | kstreith | another has summarized parts of my position much better |
| 14:53 | kstreith | can everyone please read the article at the following link... |
| 14:53 | kstreith | http://producingoss.com/en/setting-tone.html#avoid-private-discussions |
| 14:54 | kstreith | thanks, done. |
| 14:54 | gwelch | any other thought from anyone? |
| 14:54 | tclarke | one final... |
| 14:55 | tclarke | I don't think that the link above capture our situation... |
| 14:55 | tclarke | we are not proposing private discussions... |
| 14:55 | tclarke | just discussions in JIRA and IRC instead of always on the mailing list |
| 14:55 | tclarke | . |
| 14:55 | dsulgrov | also, about the link... |
| 14:55 | dsulgrov | not all JIRA issues require a discussion... |
| 14:55 | kstreith | tclarke, depends on whether you view the threshold for getting into JIRA and setting up subscriptions as effectively making things private |
| 14:56 | dsulgrov | fixing a typo in an error message does not require any discussion |
| 14:56 | dsulgrov | done |
| 14:56 | gwelch | We're coming up on an hour, are we ready to vote? |
| 14:56 | kstreith | just to poke the bear, we can also put all our discussion on webpages we provide no links to |
| 14:56 | kstreith | it wouldn't strictly speaking be private, from a HTTP standpoint |
| 14:57 | kstreith | but I think most people might consider it private from a practical standpoint |
| 14:57 | tclarke | I believe I have made all my points. |
| 14:57 | gwelch | Is there a desire to include an option in our vote to add an extra “Commit Privileges†mailing list that would house issues not of interest outside the core? |
| 14:57 | kstreith | to summarize, i believe we have only discussed the "all" or "some" |
| 14:58 | kstreith | if we vote for "some", there is still further discussion that would need to happen |
| 14:58 | kstreith | done. |
| 14:58 | dsulgrov | mconsidi? |
| 14:58 | tclarke | gwelch: that should be another discussion for another time as there are multiple ways to address that issue which are not really relevant to the current question. |
| 14:58 | gwelch | understood. |
| 14:58 | mconsidi | I think that I am willing to get a data point. |
| 14:59 | tclarke | sounds like we are ready for a vote then? any objections? |
| 14:59 | gwelch | it seems we may be ready to vote |
| 14:59 | tclarke | The vote question is: "Should we require discussion solicitation for all issues which will go into the trunk?" Valid answers are +1, -1, or 0/present/abstain ... |
| 14:59 | gwelch | Vote “1†for posting all JIRA issues on a mailing list . . . |
| 15:00 | gwelch | Vote “-1†for posting “significant†JIRA issues on a list with RSS enabled |
| 15:00 | tclarke | only those present may vote (we'll deal with rforehan if necessary later...let's assume abstain for now) |
| 15:00 | tclarke | . |
| 15:00 | tclarke | per the agenda, that's the question we are voting on. |
| 15:00 | gwelch | otherwise, abstain is 0 |
| 15:00 | tclarke | gwelch: we're not voting on two issues right now |
| 15:01 | gwelch | tclarke, are you for changing the second option to Vote “-1†for posting “significant†JIRA issues on a list |
| 15:01 | tclarke | there are no options |
| 15:02 | tclarke | it's a yes/no vote not multiple choise |
| 15:02 | gwelch | i think we may be saying the same thing, but from a differnt perspective |
| 15:03 | tclarke | the quoted question I posted is the one in the agenda...that's the question we are voting on |
| 15:03 | gwelch | let the voting begin. |
| 15:04 | tclarke | -1 |
| 15:04 | raevans | -1 |
| 15:04 | kstreith | +1 |
| 15:04 | mconsidi | +1 |
| 15:04 | dsulgrov | -1 |
| 15:04 | gwelch | 0 |
| 15:04 | tjohnson | -1 |
| 15:04 | goffena | -1 |
| 15:04 | tclarke | rforehan has left so he is 0 |
| 15:05 | dadkins | 0 |
| 15:05 | tclarke | all votes appear to be in...voting is closed |
| 15:05 | gwelch | agreed |
| 15:05 | * | dadkins is a last-minute pollster |
| 15:05 | tclarke | (2 yea, 5 nea, 3 abstain) |
| 15:06 | gwelch | the consensus seems to be to not include all issues |
| 15:06 | gwelch | anything else before we end? |
| 15:08 | tclarke | there was another question... |
| 15:08 | tclarke | on the agenda but I believe it was contingent on a yes vote |
| 15:08 | tclarke | so I move to ignore it as OBE |
| 15:08 | tclarke | . |
| 15:09 | tclarke | we'll consider this the summary...there will be a regularly scheduled meeting December 29 |
| 15:09 | tclarke | since it is the week between Christmas and New Years |
| 15:09 | tclarke | it may be cancelled if not may people are here |
| 15:09 | goffena | what are vacation schecules |
| 15:09 | goffena | ? |
| 15:09 | tclarke | feel free to drop an email to the list if you care one way or another |
| 15:09 | tclarke | I belive many of the core developers will be available on that day |
| 15:10 | goffena | ok |
| 15:10 | tclarke | I'll consider the meeting over |
| *** | Closed channel log for #opticks at 12/18/2008 3:10:16 PM |