December 15, 2008 Meeting Notes

Meeting on December 15, 2008

Agenda

  1. Attendance
  2. Announce agenda
  3. Approve December 1, 2008 Meeting Notes meeting notes. Ask for review and address any problems seen in the notes.
  4. Discuss hosting a Google Summer of Code intern for the summer of 2009. Ideas on the Google Summer of Code page.
  5. Discuss removal of subversion sandboxes and reinstatement of commit emails
  6. Should all new features be discussed on the email list? (discussion and vote)
  7. Open questions
  8. Summary and closing

Notes

Attendance

  • rforehan
  • raevans
  • kstreith
  • mconsidi
  • tclarke
  • dsulgrov
  • tjognson
  • goffena
  • gmartin_cn

Summary

The previous meeting minutes were reviewed and approved.

Opticks participation in the 2009 Google Summer of Code was discussed. Those present seemed to think this was a good idea. There was some concern about ensuring we understand all the steps and resources required. An information decision was make to brainstorm ideas and seek further details when the official announcement was made in March.

There was discussion on removing the hidden sandbox in subversion. Details on the reasons for the sandbox were presented as well as alternative solutions to these requirements. A vote will be held at the next regularly scheduled meeting to decide if the sandbox should be removed. tclarke will send an email to dev@opticks.ballforge.net with information on the alternatives so that users of the sandbox can evaluate them.

Much discussion was had about the new features discussion thread. Three main questions were discussed and a forth related question was presented.

  1. Should core developers push info about new feature development to an email list?
  2. What is the definition of new feature?
  3. What method should be used to push this information (automatic or manual)?

There was a vote on the first two questions. The results are below. The second question did not receive a quorum and a special meeting will be scheduled to finalize discussion. It will be on Wednesday [December 17, 2008|../display/opticksChat/2008/12/17/December+17%2C+2008+Meeting+Notes] at 2:00pm EST.

The final question raised, with discussion on the next regularly scheduled meeting is if the enhancement JIRA issue type is still necessary and if not, should it be removed?

Decisions

  1. Should core developers push info about new feature development (defn of new feature will be the second vote) to an email list? (9 attending, 8 yea, 0 nea, 1 abstention)
  2. Should we require discussion solicitation for all issues which will go into the trunk? (A -1 will indicate only Features require discussions) (9 attending, 1 yea, 3 nea, 5 abstentions) There was not a quorum so the issue has been tabled.

Logs

  *** Opened channel log for #opticks at 12/15/2008 2:00:27 PM
14:00 tclarke logging is now enabled
14:00 rforehan Here
14:00 raevans here
14:00 kstreith Here
14:00 mconsidi Here
14:00 tclarke here
14:00 dsulgrov Here
14:00 tjohnson Here
14:00 goffena here
14:01 tclarke todays agenda
14:01 tclarke # Attendance
14:01 tclarke # Announce agenda
14:01 tclarke #
14:01 tclarke Approve December 1, 2008 meeting notes. Ask for review and address any problems seen in the notes.
14:01 tclarke #
14:01 tclarke Discuss hosting a Google Summer of Code intern for the summer of 2009. Ideas on the SummerOfCode2009 page.
14:01 tclarke # Discuss removal of subversion sandboxes and reinstatement of commit emails
14:01 tclarke # Should all new features be discussed on the email list? (discussion and vote)
14:01 tclarke # Open questions
14:01 tclarke # Summary and closing
14:01 tclarke anyone have problems with https://opticks.balldayton.com/cometwiki/1Dec2008 ?
14:01 gmartin_cn here (sorry, a bit late)
14:02 tclarke ok, last meeting's minutes are approved
14:02 tclarke our first real item has to do with the 2009 Google summer of code
14:02 tclarke for those who are not familiar, Google sponsors a number of interns for the summer
14:02 tclarke they are tasked with working on open source projects
14:03 tclarke sponsors provide ideas, mentoring, etc.
14:03 kstreith http://code.google.com/soc/2008/ for details on last years event
14:03 tclarke google pays the interns a stipend and sponsors get a small stipend so there is not direct cost if opticks sponsors a student
14:04 tclarke we did not apply in 2008 as we did not feel opticks development resources were in place for a 3 month intern to get up to speed and do something useful
14:04 tclarke I feel we are just about ready for something like this and can be ready for summer 2009
14:04 tclarke thoughts?
14:05 tclarke we would also need to come up with some project ideas for a student
14:05 gmartin_cn From a community standpoint, is part of the goal to get Google to see how Opticks might fit into their Google Earth/Maps product space?
14:05 tclarke so consider this a brainstorming session as well
14:05 tclarke the goal from Google's standpoint is simply to get students working open source
14:06 kstreith the goal from our standpoint is to get new contributors
14:06 tclarke they also use it indirectly to search for talent
14:06 tclarke our goal is as kstreith mentioned
14:06 kstreith the actual task isn't all that important to either google or us
14:06 gmartin_cn Right, but they usually also have an ulterior motive, and it never hurts to get a company like Google looking at how Opticks can possibly fit into their plans
14:08 tclarke I think their main secondary goal is to find new employees but there's always a chance they find a product that interests them
14:08 tclarke if google gets interested in Opticks because of this, it's good for us but I'm not going to count on it
14:08 gmartin_cn In tougher economic times, even companies like Google look for some possible fit with their interests... the days of 'community goodwill' only are limited... not saying that is necessarily the case here, just that we need to think of this in terms of how we can best put Opticks in front of more influential people... and target groups like the academic space (who usually follows Google's lead)
14:09 gmartin_cn I'm not against more contributors, just looking to balance the contributor aspect with using Google to grow the outward community for Opticks
14:09 kstreith sounds like gmartin_cn is behind the idea
14:09 kstreith what do others think?
14:10 rforehan What's not to like - "free" labor and more exposure. What are the downsides?
14:10 tclarke the downsides:
14:10 kstreith not really any
14:10 gmartin_cn Well, except for coming up with appropriate tasks..
14:10 tclarke we need to provide a mentor which taps some resources, but minimal
14:11 gmartin_cn Something that is self-contained enough to be done in 3 months
14:11 tclarke if we don't have our act together, we can generate bad press
14:11 gmartin_cn tclarke: exactly..
14:11 tclarke I see that as the main reason we skipped it this past summer
14:11 kstreith true enough
14:12 rforehan So we need a well-scoped task that can be completed in less than 3 months.
14:12 gmartin_cn Keep something else in mind - the current contribution model (as necessitated by your legal folks) could cause Google some angst...
14:12 tclarke I think we are much better positioned now...especially if the task is to create a plug-in with a specific function and doesn't require much or any core moficiations
14:12 kstreith plus we didn't start thinking about it until too late
14:12 kstreith the application occurs in early March
14:12 tclarke google does not require the sponsors to use the contributed code
14:12 kstreith so we still have the month of January and Feb to come up with the ideas that students would work on
14:12 tclarke so I don't think they will get too bent out of shape about that
14:13 dsulgrov has anyone identified the things that need to happen before it would begin?
14:13 gmartin_cn tclarke: Sure, but bad press could follow (not to mention the intern feeling like they couldn't 'contribute' something if it doesn't get used)
14:14 tclarke gmartic_cn: I agree to a degree...we would plan on using the code...my point is that if it took us some time to integrate with the main trunk I don't think it would be too much of a problem...especially if the intern had a way of distributing unofficially
14:14 tclarke via COAN, for example
14:15 tclarke dsulgrov: we have not generated an extensive list
14:15 tclarke some info we don't have yet and won't until the applications open up
14:15 rforehan Haven't we had summer interns before? Any lessions learned that would apply here?
14:15 kstreith my thought is that we should put together a list of ideas that some could work on in 3 months or less
14:15 tclarke but we need ideas and a mentor...as well as Ball approval if the mentor is a Ball employee
14:15 gmartin_cn tclarke: Sure - COAN would work..
14:15 kstreith that way we have small ideas that any new contributor could take on
14:16 tclarke I've set up a wiki page for some ideas
14:16 tclarke https://opticks.balldayton.com/cometwiki/SummerOfCode2009
14:16 kstreith and it also helps us if we decide to apply for summer of code
14:16 kstreith my point i guess, is that generating a list of ideas at this point doesn't really have a downside
14:16 tclarke it's currently editable by people with accounts on the wiki only...it'll move to the front facing wiki once kstreith gets it set up (in progress)
14:16 kstreith even if we decide at a later point to skip this year's summer of code
14:17 tclarke as an aside (only partly open source applicable), Ball provides summer interns and such a list could double as an idea list for a Ball intern if necessary
14:18 tclarke or for other community members looking to get more involved and are looking for ideas
14:18 tjohnson It sounds like we have little to lose, so we should proceed with looking further into it and making preparations
14:19 tclarke I agree
14:19 tclarke anything more to say?
14:19 tclarke if not, let's move on
14:19 tclarke ok, next item
14:19 tclarke removal of subversion sandboxes and reinstatement of commit emails
14:20 tclarke we've had the /sandbox portion of subversion for a while
14:20 tclarke it's restricted to core committers and is intended for a few things:
14:20 kstreith by restricted he means only visible to core committers
14:20 tclarke development of long term branches that will span multiple releases
14:21 tclarke development of branches by ball personal which have not received ITAR release approval
14:21 tclarke with the possible extension to non-ball personal....this has not happened yet
14:21 tclarke and storage of branches from /branches which were not included in a rollup
14:22 tclarke the first couple of scenarios can be met either by working entirely in a working copy (no intermediate checkins until includion in a release)
14:22 tclarke or by using a distributed VCS to create a local branch...this works with SVK, Mercurial, and Git (probably with others, but these are the systems we have tested)
14:23 tclarke the third item (moving from /branches) would need to be addressed some other way
14:23 kstreith by we he means tclarke and kstreith
14:23 tclarke if we got rid of the sandbox, we could re-enable the commit email list
14:23 tclarke which sends an email whenever there is a commit to the repo
14:24 tclarke it also simplifies the permission scheme in subversion
14:24 tclarke thoughts?
14:24 kstreith tclarke is suggesting that we no longer have a "hidden" sandbox
14:24 kstreith and that if we still want one, it should be "public" just the same as trunk or branches
14:25 mconsidi I'm ok with that.
14:25 dsulgrov how much do the current committers use the sandbox?
14:26 rforehan I've been using the sandbox for IR&D development - I need to move that code before the box goes public.
14:26 tjohnson I make moderate use of it
14:26 kstreith rforehan: we wouldn't make the existing sandbox public
14:26 tclarke as a follow up, could current uses be moved to a public sandbox or local branches?
14:26 kstreith we would just stop using it and svn delete it from the HEAD revision
14:27 tjohnson For instance I have a current branch in it with upgrades to the Subject/Observer code
14:27 kstreith it would remain in the repo for historical purposes and would remain hidden
14:27 tclarke we would set /sanbox to read-only for core committers and r/w for admin only
14:27 tclarke and eventually delete it from the current subversion version
14:28 tjohnson the primary purpose of it was to allow branches to be worked on before getting Ball ITAR approval
14:28 tclarke correct
14:28 tjohnson is this no longer an issue - i.e. is Ball sufficiently responsive?
14:28 tclarke correct
14:28 tclarke but this does not happen very often
14:29 kstreith and as tclarke stated, you can leave the changes on your local computer, i.e. your working copy
14:29 tclarke these days, we don't move to implementing until there's itar approval except in a few rare cases
14:29 dsulgrov does the benefit of having the sandbox outweigh the benefit of having a commit mailing list?
14:29 tclarke when there's a day or two delay, most of us start work in a working copy and just don't commit until there's approcal
14:29 kstreith or you can use a distributed version control system to interface with subversion and still work it on your computer while maintaining version history just on your local computer
14:30 tjohnson I have never used a distributed VCS, so I don't know what's involved
14:30 tclarke dsulgrov: that's part of it...the commit email list is just part of the problem...it also restricts who can admin the server from collabnet's end...we could possibly relax that in the future w/o a restricted area
14:30 tclarke svk is pretty easy to use
14:31 tclarke you run a couple of commands to create a mirror
14:31 rforehan Are you looking at just the open source repo?
14:31 tclarke the creation of a branch, checkout, etc. commands are the same as subversion
14:31 kstreith rforehan: yes
14:31 tclarke in fact, svk usees subversion for storing the mirror so you can point Tortoise at the local repo
14:31 tclarke and just use svk for updating the mirror
14:32 kstreith we can defer a decision until next IRC meeting
14:32 tclarke yes
14:32 kstreith i just think trevor wanted to get the ball rolling in this meeting
14:32 tclarke this is just to get people thinking about it and start a discussion
14:32 kstreith and get people thinking about the problem
14:32 tjohnson I'll try to give SVK a look before then
14:32 tjohnson It sounds like there is no schedule driver on this
14:32 tclarke let's stop discussion now...I'll post a summary to dev@opticks along with a little info on options
14:33 tclarke we'll make a descision at the next meeting or even one after that if people need more time
14:33 tclarke anything further?
14:33 kstreith tjohnson: correct, there is no pressing schedule driver on this issue
14:34 tclarke ok, let's move on...I'll take an action item to post more info on git, mercurial, and svk to the mailing list
14:34 tclarke the next item is a discussion and vote...so this is the last chance to mark attendance if you have not already done so
14:34 tclarke any late arrivals?
14:35 tclarke ok, this is a follow up to the last meeting's discussion about new features
14:35 tclarke to summarize, should new features (and their api/functionality changes) always be discussed on dev@opticks or should we assume people ready JIRA and only discuss when further information is needed
14:35 tclarke before we resume discussion
14:36 tclarke I'll try and summarize the points on both sides
14:36 tclarke discussion of all features on the mailing list can take extra time and drives up traffic on the email list possibly causing poeple to ignore the list more often
14:37 tclarke not everyone is on IRC all the time and does not always check JIRA (mailing list acts as a "push" of information)
14:37 tclarke I'll remind people that JIRA has RSS feed support so you can subscribe to specific filters and issues and you can add yourself to a watch list for issues
14:37 tclarke which sends an email whenever that issue changes
14:38 tclarke both of which generate data pushes (RSS is technically a pull but most readers automatically poll at regular intervals)
14:38 tclarke the downside to inadequate discussion are obvious I believe
14:38 tclarke lets open up to further discussion
14:39 kstreith ok, i think we should push all discussion of new features to the mailing list
14:39 kstreith here's why
14:39 gmartin_cn Open Source communities (like the linux kernel) rely on open and transparent communications - generally the most expedient is the email list - some folks still won't use RSS (even though that is a better method, IMHO)... but discussion has to happen somewhere like an email list.. - maybe short announcements on the mailing list
14:39 tjohnson Last week there were only 11 messages on the mailing list - that doesn't seem like enough to drive people away
14:40 kstreith during 4.3.X development is the time when committers will be the most open to suggestions by others
14:40 kstreith because we know if somebody says, you shouldn't have called this method that name
14:40 kstreith we might listen and change it during a code review
14:41 kstreith but we won't change it after 4.3.0 went into binary lockdown
14:41 kstreith regardless of the right the reasons might be
14:41 kstreith so, if we're opening to listening to feedback now
14:41 kstreith we should be as "open" as possible
14:42 tclarke here's my biggest concern with always posting:
14:42 kstreith and I think the mailing list meets that criteria better than JIRA or RSS feeds
14:42 tclarke it may discourage developers from duplicating/summarizing in the JIRA issue which is the problem location for design decisions, etc.
14:42 tclarke I have a technical question for kstreith
14:42 kstreith go ahead
14:43 tclarke can JIRA be easily configured to email dev@opticks when new features more into certain states? (design review, code review, CCB review, etc?)
14:43 tclarke obviously it would be for open issues (OPTICKS-????) only
14:43 tclarke not OPTICKSPRIVATE
14:44 kstreith possibly, with a little configuration
14:44 rforehan tclarke: can't the e-mail thread be attached to the JIRA issue to maintain a log of discussions?
14:44 kstreith tclarke: i think putting the information in JIRA is just a discipline thing
14:44 tclarke I'd suggest that we try and set it up to automatically send emails to the mailing list which should automate these mailings
14:45 tclarke it also encourages people to look at the actual JIRA issue since the email is coming from JIRA
14:45 kstreith rforehan: that would be possible as well
14:45 tclarke the developer can then summary or attach as necessary
14:45 tclarke to document reasoning
14:45 tclarke thoughts?
14:45 kstreith tclarke: the e-mails that come from JIRA won't have useful reply to's set
14:46 kstreith so that may confuse more than it helps
14:46 tclarke the email list serv
14:46 tclarke will automatically change that
14:46 tclarke to dev@opticks
14:46 dsulgrov what's the primary purpose for sending info to the dev mailing list?
14:46 tclarke to encourage review by non-core developers
14:46 dsulgrov is it for information purposes or to invoke discussion on the issue?
14:47 tclarke so they can feedback about the design
14:47 kstreith what I stated earlier, "to get as much feedback from others as pssible"
14:47 gmartin_cn dsulgrov: Should be both
14:47 dsulgrov it seems like this has the potential to lengthen development time for any given issue
14:47 kstreith let me attempt to summarize dsulgrov's concern
14:48 tclarke yes...but not getting feeback can length long term maintainance
14:48 tclarke lengthen that is
14:48 kstreith his concern is that people will keep adding feedback without any realization of the time/effort they adding onto to finishing the task
14:48 tclarke this is why I suggested automatic emails at points in the process where peer review occurs since this causes a delay anyway
14:49 kstreith i think peer review is too late
14:49 tclarke design review
14:49 tclarke and peer review
14:49 rforehan Object is to insure mod's meet needs and not to micromanage how code is written, right?
14:49 tclarke there are 2 peer review stages
14:49 kstreith my answer to dsulgrov is always, "meritocracy"
14:49 kstreith http://www.merriam-webster.com/dictionary/meritocracy
14:49 gmartin_cn dsulgov: It comes down to this - if you guys want to run this project as an Open Source effort, you have to put up with some perceived 'inefficiencies' introduced by the transparency brought on by the Open Source methodology - usually it is up to the group of core committers to 'moderate' the amount of discussion and make a decision at some point - think 'benevolent dictator' like the Linus Torvalds on the kernel list.
14:50 gmartin_cn And meritocracy as kstreith points out...
14:50 kstreith exactly, at some point we are allowed to say, we're writing the code, thanks for the input, here is the decision i'm making and i'm now moving forward
14:50 gmartin_cn But even in the most perfect of communities, sometimes the leaders have to put a stake in the ground and move on
14:51 gmartin_cn But that doesn't mean that healthy and open discussion shouldn't happen..
14:51 kstreith precisely
14:51 tjohnson Having been at the nexus of one such discussion (the animation toolbar upgrades discussion), I can attest to the fact that discussing it on the dev list certainly added to the time before code was being written, but the result was much, much better for it. The animation toolbar was a source of constant complaint and the upgrades silenced that almost completely. The feedback was essential.
14:51 gmartin_cn yup...
14:51 kstreith and i will agree that for what is percieved to be really simple 1 line changes
14:52 tclarke remember, there are 3 types of issues we track....bugs, enhancements (minor extensions/features) and full blown features
14:52 tclarke we need to clarify which we are talking about if we decide to always post to the list
14:52 kstreith we can simply post to the mailing as informational
14:53 kstreith and wait until someone else turns it into a discussion
14:53 kstreith not every feature needs to be posted to the mailing as a question on how to proceed, only the complicated stuff needs to be presented that
14:53 gmartin_cn kstreith: That's generally how it works in the rest of the OSS world
14:53 gmartin_cn Correct
14:53 tclarke I don't think is a question of whether we should solicit discussion on some issues...the question is if the developer should decide which need discussion on the mailing list and which are trivial
14:54 gmartin_cn yes, exactly
14:54 tclarke my problem is with automatically posting all issues
14:54 kstreith i say the core developer cannot make that distinction
14:54 kstreith so everything should go out
14:54 tclarke not with posting issues that the developer feels need discussion
14:54 kstreith by, everything I really do mean everything
14:54 tclarke I think that posting everything when only 10% will have discussion is a problem
14:55 kstreith the reason i say the core developer can't make the distiction, is because they don't have thousands of lines of plug-in developer code at their desk
14:55 tclarke because it causes people to ignore most of the mailings
14:55 tclarke there's too much noise
14:55 goffena at what point do you start seeking input via the mailing list?
14:55 gmartin_cn If if gets to be too much email, the possibility always exists that you do 'sub-communities' of interest...
14:55 tclarke we've seen this sort of behavior in the past
14:55 tclarke were people tune out new threads
14:55 kstreith which is why I would argue that we can batch together stuff that is simple into 1 e-mail instead of 15 e-mails
14:55 goffena have you already discussed on irc with core developers and a few definite stakeholders from the development community?
14:55 tclarke because they don't want to take the time to review something they won't likely care about
14:56 kstreith so I'd rather not have JIRA auto send e-mails
14:56 tclarke I'd prefer to have that decision made by community members who actually care enough to review everything
14:56 tclarke which is why I'd rather have them visit or RSS JIRA
14:56 kstreith goffena: we don't have a good rule of thumb at the moment, that's what this discussion is trying to get at
14:56 tclarke so we don't alienate the casual email list users
14:57 tclarke or even have a bot auto-post changes to IRC
14:57 kstreith ok, to be clear i'm only suggesting we do this type of posting during heavy new feature development, i.e. something like 4.3.X
14:58 tclarke but that's a filter
14:58 gmartin_cn Is there a possibility you might split the dev list into 'core' and 'plugin'? Would that help alleviate some of the perceived 'noise?
14:58 kstreith if we are in bug fix mode, i'm ok with the other model, which says go check jira and we'll only post if it's really important and interesting
14:58 tclarke what about trivial bug fixes that happen during that time...you wouldn't post about the same fix during 4.3.1 then?
14:58 tjohnson E-mail is ubiquitous. IRC and RSS are not. The e-mail is your broadcast.
14:58 tclarke you're now setting the filter you indicated you didn't want to set
14:58 tclarke there are still interface changes, etc. in the "off" times, they just preserve binary compat
14:59 tclarke I think the real problem seems to be the scope of dev@opticks
14:59 tclarke is it there to encourage questions, etc. from plug-in developers?
14:59 tclarke or to facilitate core development
15:00 kstreith keep going, i'm interested
15:00 tclarke we've been using it for both, but I'm not convinced that's what it should be
15:00 tjohnson why is this an either/or?
15:00 kstreith we have been using it for both
15:00 gmartin_cn Good point tclarke.. I'm interested too... (since I suggested a split along some lines)
15:00 tclarke since the "causual" plug-in developers will be turned off by core development discussion (including discussion of all new features)
15:00 kstreith and my thought has always been that we would split it once traffic levels dictated it
15:01 tclarke and core and plug-in developers who are more invested would care about the "chaff"
15:01 tclarke I was suggesting making the split via email/RSS
15:01 kstreith ok
15:01 tclarke but perhaps two emal lists would better serve
15:01 kstreith i will say that even if we split the list into two, for this cycle in particular, i would push to move the traffic onto the plug-in dev oriented on
15:02 tclarke I'd love all people on dev@opticks to read these designs (as individual posts or digests)
15:02 gmartin_cn Have there actually been complaints about the signal to noise ratio, or is that just a perceived issue?
15:02 tclarke but a lot of developers don't even read the release notes
15:02 kstreith because i'm trying to engage our current plug-in devs to become interested
15:02 tclarke how can you expect them not to tune out all that stuff they will likely see as not interesting
15:02 tclarke gmartin_cn: not direct complaints
15:02 kstreith gmartin_cn: i have not heard of complaints, can't speak for others
15:03 tclarke I have talked to developers about issues posted to the mailing list
15:03 tjohnson are you expecting a vast increase in the level of traffic? The last three weeks were 11, 29 and 9 e-mails respectively on the dev list.
15:03 tclarke and they have told me.."Oh, I stopped paying attention to that"
15:03 tclarke tjohnson: it's not so much the absolute volume, it's the s/n
15:04 tclarke I realize it's kind of stupid to not read a half dozen or so emails a week, but people do it
15:04 kstreith alright, let
15:04 tclarke maybe the solution is to chalk those people up as hopeless when it comes to reading emails
15:04 tjohnson perhaps we could do something to better mark the threads of discussion?
15:05 tclarke that was part of what I was hoping to get out of JIRA auto-emails
15:05 kstreith my suggestion is we title all 4.3.X messages as such
15:06 kstreith i think part of the problem is that core committers don't think we will get any feedback
15:06 kstreith but that's NOT a reason to not engage others
15:06 kstreith you'll never get feedback if you don't create an environment for feedback
15:07 kstreith ya know, "build it and they will come"
15:07 tclarke ok, sounds like we need a couple of votes
15:07 tclarke first: should we push info about all features
15:07 tclarke second: what is the delimiter re: when to stop...is it a dev line delimiter? or a JIRA issue type delimiter
15:08 kstreith plus there is benefit to posting and people just reading, but not responding
15:08 tclarke third: what method should we use to push (automatic JIRA email, manual emails, or digest emails?)
15:08 kstreith now they are aware and can plan ahead for their plug-in development
15:08 kstreith and they might start to get excited about the next release before it even comes out
15:08 tclarke do those should like the main questions we are trying to answer?
15:09 rforehan What's the scope? every chance in code, enhancements, added features, etc?
15:09 rforehan chance->change
15:09 kstreith i'd like to get more dicussion from others before we push for a vote
15:09 tclarke that's question the second
15:10 tclarke I'm suggesting we have the first vote...depending on the outcome, we may not need further discussion
15:10 tclarke unless people feel the second two questions will strongly impact their vote for the first
15:11 kstreith ok, let's vote on the first question
15:11 kstreith just to see where everyone is
15:11 tclarke the vote is: "Should core developers push info about new feature development (defn of new feature will be the second vote) to an email list?"
15:11 tclarke please vote +1/-1 or present
15:12 kstreith +1
15:12 tclarke +1
15:12 raevans +1
15:12 tjohnson +1
15:12 gmartin_cn +0 (present)
15:12 mconsidi +1
15:12 goffena +1
15:12 rforehan +1
15:12 tclarke ok, that's enough to procede
15:12 dsulgrov +1
15:12 tclarke now, what does feature mean....
15:13 tclarke all JIRA issues? just certain types (feature, enhancement, bug)? Just certain times in the dev cycle?
15:13 tclarke this is for required/automatic postings
15:13 tclarke you can always manually post about any jira issue
15:14 tclarke I think only features should be required but at anytime in development...the other two have a very focused nature
15:14 tclarke any perceived problems are already being discussed by developers on those two types
15:14 kstreith i think for this 2 month dev cycle, it should be everything
15:14 tclarke I think that would cut down the s/n and push only the major impact issues
15:14 kstreith an 2 month experiment, we'll call it
15:15 tjohnson "Major" is in the eye of the beholder. I would suggest all enhancements and new features.
15:15 kstreith and after that, we should talk about the benefits of that approach and what we should do after that
15:15 tclarke I would argue that anything needing that much design feedback should be a feature
15:15 tclarke if it's an enhancement and has that much impact, it's mis-classified
15:15 dsulgrov wait
15:15 dsulgrov are we talking about only issues that need design feedback?
15:16 tclarke I'm talking about anything marked as "Feature" in JIRA
15:16 dsulgrov there are a number of enhancements that do not require design feedback (e.g. OPTICKS-294)
15:16 tclarke kstreith is talking about any change made to the core
15:16 kstreith alright, let's get to concerte examples
15:16 rforehan "Major" is a relative term - look at how long some posts are just on the location or caption of a GUI button.
15:16 kstreith tclarke: correct
15:16 kstreith we might want to change the name of an enum because it conflicts with Hdf
15:16 tclarke if we will require that much design and feedback on all issues, we should get rid of the Enhancement category
15:17 kstreith that seems easy, no need to talk to anyone about that
15:17 kstreith except that we pick a name that conflicts with another plug-in dev that they use 1000 places
15:17 kstreith oops
15:17 tclarke I say we can only go so far
15:17 kstreith that's my point, we can't know what is "major"
15:17 tclarke we should be namespacing but we are not...but plug-in devs should also be namespacing
15:17 kstreith but that seems simple, so put it an single e-mail that mentions 10 other things
15:18 tclarke we can only account for our own past design mistakes, not theirs
15:18 tclarke I think that's even worse than individual emails
15:18 kstreith except maybe the conflict isn't in their code
15:18 tclarke people are even more likely to miss it
15:18 kstreith maybe it's in a external module they are using
15:18 tclarke ok, so our design should add a namespace
15:18 kstreith so on and so on
15:19 kstreith i'm just asking that we try this approach for this dev cycle and this cycle only
15:19 tclarke if we are not able to /allowed to make a decision on the "size" of the feature, we should not have enhancements
15:19 kstreith after that we can hold a lessons learned on the mailing list
15:19 kstreith and see what everyone thinks
15:19 tclarke which is ok if we decide that, but lets remove the option
15:20 kstreith tclarke: as time moves forward, i've been thinking the same thing
15:20 tclarke also, your earlier response re: only making this mandatory during trunk/future dev (although you may have changed your mind by now)
15:21 tclarke is that we can't limit it to just that time by your argument that we don't know when the impact is high
15:21 kstreith ok, i've simplified my request
15:21 kstreith if it goes into trunk/future can we always talk about it on the mailing list, until we create trunk/4.3.0?
15:22 kstreith after that, we'll talk about the whole experiment in another IRC meeting
15:22 kstreith and decide whether it was good, bad, etc.
15:22 dsulgrov including bugs?
15:22 kstreith correct
15:22 kstreith if the code gets commited to trunk/future, then talk about it on the mailing list
15:23 tclarke no...if you want to always talk about it, it should be always not just during future...that doesn't mean we can't re-evaluate later but the rule should always hold until repealed
15:23 tclarke if you want to set the revist to 4.3.1 timeline, that's fine
15:23 tclarke the difference is how we think about what is imporant
15:23 tclarke looks like we've got two main methods of deciding what to post: Every issue, only issues marked as Feature (both carry the possibility that Enhancement goes away as a type)
15:24 tclarke I would like to conduct the vote on that question if everyone thinks they have enough information
15:24 dsulgrov I am not in favor of removing Enhancement issues
15:25 tclarke that's a separate question
15:25 tclarke we can discuss the possibility of removing it later
15:25 kstreith yeah, that's a discussion for next IRC meeting
15:25 kstreith if you want to talk about that is
15:26 tclarke the vote is: "Should we require discussion solicitation for all issues which will go into the trunk?" +1/-1/present A -1 will indicate only Features require discussions with the Enahncement question to follow at another meeting
15:26 tclarke -1
15:26 dsulgrov I think we are trying to come to a resolution too fast because this has implications on how development occurs
15:26 *** goffena (i=0cbc9d81@gateway/web/ajax/mibbit.com/x-11951dec07d2ed75) has quit IRC [Read error: 104 (Connection reset by peer)]
15:27 tclarke then vote present
15:27 tclarke if there's not a quorum the question will be tabled
15:27 kstreith +1, but I don't think that suprises anyone
15:27 raevans -1
15:27 tjohnson present
15:27 rforehan present
15:27 dsulgrov -1
15:27 mconsidi present
15:28 tclarke I think we are missing two votes...they will be abstentions if left blank
15:28 kstreith correct, missing goffena and gmartin_cn
15:28 tclarke looks like goffena dropped
15:29 gmartin_cn +0
15:29 gmartin_cn Sorry..
15:29 tclarke I'll wait a minute for him to return
15:29 tclarke so far it looks like there's no quorum
15:29 tclarke ok, I'll assume goffena is not returning right now and his vote will be a +0
15:30 tclarke there's not a quorum so we'll put this off for the next meeting...reply to the meeting minutes email with offline discussion
15:30 tclarke or we can continue during the open discussion period if poeple want to continue now
15:30 tclarke I don't want to drag on too long though
15:31 tclarke ok...are there any other topics for discussion this week?
15:31 tclarke ok, I'll put together the meeting minutes including the action item I have picked up
15:32 tclarke we have decided to post all new "features" for discussion to the mailing list
15:32 tclarke since we have not clarified the meaning, I'd suggest we including any new trunk/future discussions at least until the next meeting
15:32 *** tjohnson (i=0cbc9d81@gateway/web/ajax/mibbit.com/x-c6b6c46ad705d946) has quit IRC ["http://www.mibbit.com ajax IRC Client"]
15:32 tclarke we won't vote on that, I'll assume silence is agreement
15:33 dsulgrov please clarify your statement
15:33 tclarke any new development in trunk/future (new branches) between now and the next meeting should be brought up on the mailing list until we can hold a vote about relaxing this requirement
15:35 tclarke do we want to schedule a special meeting later this week just to discuss/vote on this issue?
15:35 tclarke or are we ok waiting two weeks?
15:36 tclarke I don't want to push this meeting out any further but I also don't want to miss any discussion if the vote decides to discuss everything
15:36 tclarke since we havn't voted on the second question, the above would be a suggestion to core developers
15:37 kstreith if we don't have consensus, we might want to add a meeting in order to get consensus
15:37 tclarke ok, I'll schedule something later this week
15:37 kstreith i'd rather not have this hanging out there for 2 weeks
15:37 tclarke any suggestions on times?
15:38 tclarke how about wed afternoon...2:00 EST
15:38 tclarke that gives us a couple of days to digest
15:38 tclarke and we'll vote again just on this issue
15:38 kstreith sounds good for me
15:38 tclarke until then, things will continue as they have been
15:39 tclarke ok...meeting will close in 1 minute
15:39 tclarke ok, meeting is closed
  *** Closed channel log for #opticks at 12/15/2008 3:39:53 PM

Labels

meetings meetings Delete
Enter labels to add to this page:
Please wait 
Looking for a label? Just start typing.