PMC Functional Meeting 20100408
Date: 2010-04-08
Time: 9PM GMT
Venue: irc #adempiere-team
Support Spreadsheet: Adempiere PMC Functional
Chat times in GMT-5
(16:00:34) ar_howard [~quassel@ppp118-208-145-17.lns20.bne1.internode.on.net] entered the channel.
(16:01:38) CarlosRuiz: Hi Allan
(16:02:43) ar_howard: Hi Carlos
(16:03:00) ar_howard: Steven is in NZ yesterday and today, so I don't expect to see him here
(16:04:51) CarlosRuiz: for the agenda today I would like to discuss the 3 points you raised past meeting - let me check if Joel is available ...
(16:06:21) JoelS_ [~4bd36990@gateway/web/freenode/x-yoecudfjdlqehvft] entered the channel.
(16:06:28) ar_howard: Hello Joel
(16:06:38) JoelS_: greetings functional team
(16:06:42) JoelS_: Hi Allan
(16:06:57) JoelS_: Carlos, did you want to put forth your questions?
(16:07:31) CarlosRuiz: yes - I was saying this to Allan: for the agenda today I would like to discuss the 3 points you raised past meeting
(16:08:20) CarlosRuiz: 1) What is most effective in getting some of developments from implementors in front of the community (making statement general)
(16:08:20) CarlosRuiz: 2) How to get those integrated if deemed desirable
(16:08:20) CarlosRuiz: 3) if not desirable the best way to keep other integrations in the public eye in case someone else wants to adopt in the future
(16:08:48) ar_howard: Should we start with (3) and work towards the higher goals?
(16:09:32) JoelS_: on #1- our current channels * SourceForge and * wiki are too hard to follow
(16:10:08) JoelS_: we need to create a permanent webpage- maybe with key links maintained on .com home page
(16:10:34) JoelS_: we can proselytize in the forums that they are there
(16:10:48) JoelS_: someone needs to own this process- or maybe our group does
(16:10:49) ar_howard: Is this where functional team crosses over documentation - is there a possibility of a conflict?
(16:11:23) CarlosRuiz: Joel - do you mean just a list of things worked per project? or about code?
(16:11:25) JoelS_: mmm, i could see documentation group drives their need for better documentation from this list
(16:11:35) JoelS_: it's more like our way to publish what we think is good
(16:11:43) JoelS_: our=functional group
(16:12:00) JoelS_: 'advertisement' of what is being done
(16:12:23) ar_howard: I was thinking, with regard to documentation, that we need functionality in AD which more strongly guides people to the right places and also allows people to easily publish from the point of documentation usage (home I'm not getting off topic here)
(16:12:56) CarlosRuiz: a little :-)
(16:13:13) JoelS_: but this is a good related topic, let's come back to it
(16:13:33) JoelS_: right now, when I make a new thing, there is very little value in my giving it to community. no one seems to notice or care
(16:13:52) JoelS_: but if I thought someone wanted it, i would be more motivated
(16:14:13) JoelS_: so if we have good publication, maybe we can generate more interest from contributors to giving back
(16:14:17) CarlosRuiz: I'm wondering - why we as implementors want to publicize (advertise?) the work:
(16:14:17) CarlosRuiz: - to state that we want to contribute
(16:14:17) CarlosRuiz: - or to get customers because we have something unpublished they need
(16:15:07) CarlosRuiz: :-) tricky question
(16:15:30) CarlosRuiz: but what I want to point is that there are implementors that like to share - and there are implementors that like to advertise without sharing (just as a marketing channel)
(16:16:03) ar_howard: For implementors it is hard - they are paid to do a job for the customer, but not paid to push that up the trunk
(16:16:17) CarlosRuiz: I think we need to focus on the first group - those who want to share and there are some stoppers for that
(16:16:19) ar_howard: They need to build into their work a margin to allow a contribution
(16:17:10) ar_howard: Carlos - I think that is a good place to start - focus on those who want to make a contribution and make an example of them to sell the concept to others
(16:17:35) CarlosRuiz: interesting - in open source is hard to distribute the costs of development between several projects (unless you hide)
(16:17:37) JoelS_: maybe having the code/ad changes loaded to a branch is a requirement for something being listed
(16:17:58) ar_howard: What do you think are the show stoppers?
(16:18:51) CarlosRuiz: I like the idea - upload in a contributions folder whatever you have - it can be work from others to integrate in trunk - indeed many things have arrived to trunk this way
(16:19:10) CarlosRuiz: > What do you think are the show stoppers?
(16:19:10) CarlosRuiz: good question
(16:19:46) CarlosRuiz: doing the triage I've noticed how abandoned we have the contributions, feature requests, etc
(16:20:07) CarlosRuiz: I guess people wanting to contribute can get tired if things are abandoned without a single comment
(16:20:26) CarlosRuiz: that could be one stopper
(16:20:42) ar_howard: So abandon = means unmaintained? Do they become unmaintained because other functionality breaks them and people are tired of re-fixing?
(16:21:30) CarlosRuiz: no, no unmaintained, I mean, totally unattended
(16:21:54) ar_howard: OK, so they go from an idea, to partial code, and never get completed?
(16:22:01) CarlosRuiz: I have found many bugs reported - even with the solution attached - and without any comment - and I mean bugs 2 years old
(16:22:38) CarlosRuiz: so, if I contribute and nobody cares or answer - then I guess I'm discouraged to do it again
(16:23:33) ar_howard: Hmm - interesting, I thought contributing would come from someone who had a real need, and that would be the driver to complete. Is the issue that people are coming up with ideas that don't have a real customer behind them - just a "wouldn't it be nice to have"
(16:23:33) CarlosRuiz: another stopper (hard to workaround) -> sharing dictionary changes. Or you share with official dictionary - or you share with 2pack - both represents extra work commonly not charged to customer
(16:24:20) JoelS_ has left the channel (quit: Ping timeout: 248 seconds).
(16:25:09) CarlosRuiz: but I suppose we need to try it again
(16:25:26) CarlosRuiz: Norbert tried some time ago with this page:
(16:25:26) CarlosRuiz: http://www.adempiere.com/index.php/Feature_Development_Collaboration
(16:25:38) ar_howard: I think at some stage Adaxa were thinking of as part of their charge to cover costs associated contributing back - I wonder if other implementors are thinking the same?
(16:27:58) ar_howard: I suppose, if the initial contribution was supplied as a separate branch (complete with data) and there was enough interest a group could form around that seperate branch to integrate it's ideas. That would only work for things everyone wanted however
(16:28:39) ar_howard: You would have to have a "branch trimmer" come along every few months and prune off those branches that withered
(16:28:56) CarlosRuiz: > 1) What is most effective in getting some of developments from implementors in front of the community
(16:28:56) CarlosRuiz: if I shoot the question back to you - what do you need? what will make the process easier for you?
(16:30:26) ar_howard: For us - if we could simply provide a copy of production data, have it automatically de-identified and supplied with the code this would make it a whole lot easier for people to assess (and if that branch were available via the Web without needing to be installed would make it perfect)
(16:30:58) ar_howard: ie so people can try it out for themselves with minimum effort
(16:31:50) ar_howard: Of course documentation is an important part as well - which is why I get frustrated that we have a central dictionary in AD of "help" that is not helpful
(16:32:10) ***CarlosRuiz thinking ...
(16:32:47) CarlosRuiz: a "data scrambler" is doable - it must not be so hard to develop, at least for basic tables
(16:33:03) CarlosRuiz: but a "demo" server per functionality sounds hard to achieve
(16:33:36) CarlosRuiz: indeed the code will need some work to make it integrable with any branch
(16:33:41) ar_howard: Perhaps if a demo server were hard, it could be a "featured demo each month"
(16:34:40) CarlosRuiz: but let me explore more - you can publish the codes, with a scrambler you can publish a dump database (and so, you're somewhat publishing the dictionary changes needed)
(16:34:54) ar_howard: sorry, what I meant by code - take the whole code base of the customer (ignoring what was custom and what is not) for demo purposes. The integration can come later
(16:36:04) ar_howard: Also for the scramler you want a size filter (so a database becomes a manageable size)
(16:36:20) CarlosRuiz: that sounds harder - but feasible
(16:36:51) CarlosRuiz: but anyways - even without data - just the dictionary someone else could do the integration work
(16:36:51) ar_howard: Adaxa have already done a recursive delete function for whole tables - it was done for our implementation
(16:37:44) ar_howard: without the data is extremely hard - in our case a lot of stuff was put into data to allow it to be reusable in other projects. Without understanding the data you would be extremely confused
(16:38:23) ar_howard: I find that with Gardenworld now - the data is minimal and it undersells the product
(16:38:25) CarlosRuiz: yes - I see - sample data is key when you develop configurable things
(16:40:03) CarlosRuiz: anyways the most important is to publish the code + dictionary
(16:40:09) ar_howard: Yes, AD data is one thing, but customer data points to real things getting done (provides a model for others to understand the links between the technical and the real world)
(16:40:40) CarlosRuiz: sample data can be replaced with documentation - and even documentation can be constructed with reverse engineering
(16:40:57) CarlosRuiz: at least to ease contributing code + dictionary can help
(16:41:28) ar_howard: I disagree - people don't want to read loads of documentation unless they have to. Most business people would prefer to scan some data to see how it is presented and being used
(16:42:39) CarlosRuiz: Allan, what I mean is - the first step must be make it easy to publish
(16:42:39) CarlosRuiz: if we push the contributor to write extra-documentation or write extra-programs then we're keeping the stopper
(16:42:44) ar_howard: Coders think logically about the impact of something and how concept could be twisted to their own needs. The majority of others look at the data and see firlds of interest and how that could be used for them.
(16:43:20) CarlosRuiz: not saying documentation and sample data is important - but if they're hard to do - then at least start contributing code+dict
(16:43:40) CarlosRuiz: if you want to get attention to your contribution the easier is to publish some screenshots
(16:45:02) CarlosRuiz: sorry - not saying that doc and sample data is NOT important
(16:45:33) ar_howard: Screenshots are a teaser - which could bring people to particular branches to review
(16:45:38) CarlosRuiz: but if they're stoppers - then we need to try to lower the bar - make it easy to publish - and let others to complete
(16:47:46) ar_howard: The previous thinking between Adaxa and ourselves was we could provide "customization.jar", the source code of this, plus scrambled database as the initial contribution. Then perhaps functional team could pick screenshots of things they found of interest in a review and write an article about that particular version?
(16:48:56) ar_howard: And the implementors take the credit for uploading as part of their sponsorship - ie you only get to trade your name if you upload a sample branch for review?
(16:51:26) CarlosRuiz: I would say - you could start with just dictionary instead of scrambled+trimmed databse
(16:51:33) CarlosRuiz: that's to make it easier to publish
(16:52:57) ar_howard: Yes, some customers may be happy with that - but I am guessing majority won't be happy until they can easily scramble+trim (privacy laws, business specific financials, etc are too exposed)
(16:54:42) ar_howard: I know our system wouldn't work without the data - during implementation our project team thought things didn't work when in reality it was just a lack of some types of data or improperly converted data)
(16:55:45) CarlosRuiz: that can be in a posterior contribution
(16:56:28) ar_howard: Carlos, I like us starting with (3) - it has lots of potential to then allow work on (1) and (2) without wasting effort when not warranted
(16:57:20) CarlosRuiz: I would think on something like this:
(16:57:20) CarlosRuiz: minimal sharing -> code + dictionary (this is easy to share)
(16:57:20) CarlosRuiz: if you want to get attention -> consider publishing some screenshots
(16:57:20) CarlosRuiz: if you want to contribute complete things -> consider publishing sample scrambled/trimmed data / also documentation can help a lot
(16:57:55) ar_howard: Perhaps we can use us as a test case - Customer and Adaxa are willing to contribute, Joel is willing to do work to help. We seem to have the foundation stones for a (3) test case?
(16:59:52) CarlosRuiz: what is your thinking on [3]? things that you publish as [1] but don't get attention?
(17:00:43) ar_howard: My thinking is that you have to gain approval to move from (3) to (2) to (1) - it's not automatic and goes via the functional team
(17:01:04) CarlosRuiz: ah I see - we're on the same track :-)
(17:01:18) CarlosRuiz: I'm thinking on [1] the same way you're thinking on [3]
(17:02:20) ar_howard: Ah yes, we have to ascertain the popularity of things before they make it into core
(17:02:48) ar_howard: This will bring stability and functionality together
(17:03:34) CarlosRuiz: at first - somebody will need to make the integration work - but at least is published to let others to work on that
(17:04:22) ar_howard: a working example should make integration a lot easier
(17:04:58) CarlosRuiz: of course - but if that delay the sharing - I prefer to encourage the principle: release early
(17:06:01) trifon: be carefull of the caowboy Trifon publishing early thing :)
(17:06:08) trifon: things in trunk :)
(17:06:23) trifon: just joking.
(17:06:48) trifon: it is very important to release early in order to give chance other to review and join.
(17:07:07) CarlosRuiz: yes Trifon, that's something good to discuss - publish in contribution or branches, not directly in trunk.
(17:07:07) CarlosRuiz: Allan - I think the meeting time is up - I liked a lot this meeting - one of the most important things is how to encourage contributors and this chat with you is enlightening for me
(17:08:16) ar_howard: OK then, I think we are progressing. I hope Joel and Adaxa have managed to meet up (having difficulties) and we can then perhaps progress a pilot of the things we are discussing
(17:09:08) ar_howard: See you next time.
(17:09:14) CarlosRuiz: great - I'll try to coordinate next meeting with Joel and let you know
(17:09:28) CarlosRuiz: thanks a lot for attending
