Notes from the Higgins Developers
call on Thursday, June 12
Attendees
=========
Charles
Andres
Paula Austel -
IBM
* Jeff Broberg
CA
Duane Buss -
Novell
Anthony Bussani -
IBM Zurich
Greg Byrd -
NCSU/IBM
* Brian Carrol -
Serena
* Tom Doman -
Novell
* Andy Hodgkinson -
Novell
Valery
Kokhan
- Parity Ukraine
* David Kuehr-Mclaren -
IBM
Mike McIntosh -
IBM
Tony Nadalin -
IBM
Dale
Olds -
Novell
Ernst Plassmann -
IBM
Uppili Srinivasan -
Oracle
* Drummond
Reed -
Cordance
* Bruce Rich -
IBM
* Mary
Ruddy
- Meristic/SocialPhysics
* Markus Sabedello -
Parity
* Jim Sermersheim -
Novell
*George Stanchev -
Serena
Daniel
Sanders
* Paul
Trevithick -
Parity/SocialPhysics
* Brian
Walker
- Parity
Jeesmon
Jacob -
Parity
* Carl Binding -
IBM
Tom Caroll -
Parity
Ernst
Plassmann - IBM
Mohamad -
Oracle
*Hank
Malden
* Attendees
Meeting
Notes
Agenda
======
[Mary} Brian will
start with the first item, the status of 1.1M2.
1) [Brian] 1.1M2 (13-June
target)
---------------------------------
* See: http://wiki.eclipse.org/Higgins_1.1M2
* All 1.1M2 items were
to be closed by yesterday 11-June.
* Stable build planned
for 13-June.
[Brian] A couple of
updates. We have a list of tickets that we will be releasing. I sent a
status update to the list today. Anything not done today has been moved to
M3 or to backlog. I have a list of 30 bugzilla items We did pull in use of
Apache commons logging from M3 to M2. Prior to being able to release M2 we
are dealing with some issues with the daily builds. Peter and Valery have
been working with that most of today. They did make some progress, and also
are working with the Eclipse webmaster on some file access issues. We need
to get that cleaned up before we do the M2 release. There may also be some
IdAS interface issues.
[Jim] I don't know.
All of the IdAS build seems to be working this morning.
[Brian] The date isn’t
super critical. Could revert back or push it out a couple days. Unless
we get these things figured out today or in the morning, it is unlikely we will
be able to get the M2 out on Friday. Need to get daily builds cleaned up
first. So that will be the approach we will take: to push to get the daily
builds stabilized. If we need to take a different approach I will reach out
to people.
2) [Brian] Update on
configuration automatic runs of test scripts
[Mary] Next item is an
update on configuration of automatic testing. Brian has been focused on M2, so not
much has been done since last week.
[Brian] We have a call set
up today at 2:00 EDT with one of the Buckminster project
leads about what they are doing for auto testing. Tom will advise us on what
they do. Mike has graciously volunteered to be the guinea pig for his
component area. So we are focused on both component and the solution
level. We will start at the
component level. Will have more tactical updates next week. Mike any other
commentary?
3) [Brian /David]
internationalization requirements
[Mary] The next
topic, which is another infrastructure item, is to talk about
internationalization requirements.
David, it would be great if you would tell us what is needed from your
perspective.
[David] We are working on
using IdAS in some projects in IBM. We require
internationalization. Need to get
that into test. Also there is the issue of the ability of API's to process strings, particularly
for identifiers. One of the team members posted some questions. How do I contribute to help and drive
this? There is a bugzilla item for internationalizing the IdAS
component. There are other Eclipse
standards, but they don’t quite apply.
[Brian] There was some
discussion on the IRC a few days ago. You are spot on, we need to iterate
quickly on what standards we need to adopt, pulling on Eclipse standards, and
then do an assessment component by component, starting with IdAS of course, then
determine what follow-on development we need to have.
[Dave] Is that something
that people are already looking at? Or could we do a
proposal?
[Brian] That would be
great. Mary and I are happy to
shepherd the process along... I did start to post some of the Eclipse practices.
It would be great if you could start with the business requirements that you
have with IdAS.
[David]
Great.
[Brian] The links posted to the dev list or wiki
page are just a repository for the proposal. I’ll send the link to the Higgins dev
list. You can use it as a landing
page.
[Mary] David we appreciate
your volunteering to help with this.
4) [Mary] Home
page design update. We now have a prototype that meets our
project objectives http://wiki.eclipse.org/Higgins_home_page_project_plan of providing more
integrity and clarity for Higgins as a project (that is not part of
the IDE). See Higgins home
page template at http://www.eclipse.org/higgins/nursery/index.php
[Mary] A number of us have
been working on the redesign of the Higgins home page. We have discussed this on
other calls. There were a number objectives for the redesign, see the link in the agenda, and we have met them. They are to improve the integrity of
Higgins as a project, to improve the navigation so that visitors don’t get lost
in the Eclipse website, to have the home page be professionally designed and
make the project easier to new visitors to understand. We now have a new template for the home
page, which you can see at the link above, and have created some sample sub
pages. We wanted to get feedback from the dev list. Next we need to run the new
design by a group at Eclipse. [Note
that we plan to put the new design into production as soon as possible (by the
start of Catalyst)]
[Mary] Any feedback?
5) Now that we have a
new home page template, a proposal to restructure the download
WebPages has been added to the bottom of this page http://wiki.eclipse.org/Deployment_Requirements#Proposal
[Mary] Now that we have the
new website look and feel, we can resume work on planning to restructure the
download pages. The proposal is to combine the download and components pages and
to streamline the other pages.
[Mary] There is a link on
the agenda, that is a proposal for requirements for restructuring the downloads
pages.
[Brian] From my
perspective, it would be nice if we could re organize the page so the top part
is highly consumable. Make that clean and easy to consume, and put all the
gory stuff to the bottom.
[Jim] If we have a
component page that is HTML, who can edit that? In past it was more important for
everyone to be able to easily edit it as it was changing quickly.
[Mary] You are right. We wouldn’t have suggested this last
year, when we were encouraging everyone to get their new components documented.
Question, do you think it is ok to move to using a web page
now?
[Jim] It is reasonable as
long as we don't slide back into a phase where we need to update it all the
time. Might want to also put it under SVN control, then request that it be
copied over [to the web CVS].
[Mary] That makes
sense. We have persons who can
update the web site in both the European and US time zones. Other comments or
responses?
[Mary] It would be nice to
have the LED again showing which builds had run or failed. That is very helpful on days like
today.
[Paul] We really need to
figure it out. The only
modification is the LED. Maybe we could change the image file to which it
points in terms of both saving time and keeping it very
flexible.
[Mary] If the link points to an image
file and image file itself is changed….
[Jim] Actually we were
changing the image. If component x had a build status LED, at the end of
the build we would upload a green LED, and if it [the build] failed, we would
upload a red LED. So it wasn’t on the wiki page.
[Paul] The image resources
were part of the wiki resources.
[Jim] I tried pointing to
images that weren't under the purview of the wiki, but I believe what was
happening was that there was an external link arrow next to the LED light, and
that was aesthetically unpleasing. This was just a couple of hours of fooling
around. There is probably a way to make it work
better.
[Mary] So we should try one
more time to make LED’s work on the wiki before we do the
conversion.
[Jim] I was just looking in
the log for one of these failed builds. They say they can't find the IdAS
API package at all, like it didn’t
build. But it did build. Is there someone who knows how the build
package works who can look at that now?
[Brian] Peter is looking at
that now.
[Jim] We can proceed to the
next item now.
[Mary] Yes, if no-one has
other questions.
6) [Mary] ICF launch
moved to June 24 to allow more time for pre-briefings.
[Mary] We just wanted to
let people know that the Information Card Foundation launch has been moved to
the week of Catalyst, June 24, to allow more time for pre-briefing the
press.
7) [Jim]
IdAS
[Mary] If there are no more
additional topics, then Jim has the floor to talk about
IdAS.
[Jim] I sent out some links before
the call:
http://wiki.eclipse.org/IdAS_API_Extensibility , http://wiki.eclipse.org/IdAS_API_Extensibility#Extension_presets
[Jim] Last week we were talking
about using attribute values as references to entities. So the request was
it would be nice if the app developer, the IdAS consumer, could read their
attributes, if they find one that is a reference, de reference it. So that
the IdAS consumer wouldn't see a pointer or reference value come back. They
would see the de-referenced value come back. So I started identifying methods
where it would make sense to have that capability. There were dozens. You
could ask a Boolean to deference the value, but then what that means is
that all the existing context providers would need to be updated so that they
can compile. And anyone calling these methods would need to update them so that
they can compile. There is another solution to just overload
it...
[Jim] What worries me is if we
start a precedent of creating new methods. So it got me back to
thinking, maybe we could revisit the extensibility. I talked to Duane and
he suggested that we have this configurable component interface. It
is a well understood way to change or add initial setting. So his thought
was why don’t we allow the dynamic method. (I.e. put a setting in the
configuration file to de-reference attributes.) It would be easy and nice.
De-reference these attributes rather than other attributes. That would be a
great idea, but that is a configuration setting, they are typically read from
some configuration file, they are not so dynamic. I don’t know if it is
even possible to re-configure a component. I wish Mike was on the call. If you
call configuration a second time, does it re-initialize?
[Paul] It is interesting. Google was
looking at our system. One of the things they didn't like was it [configuration]
was static. They moved to a dynamic API. I see where you are
headed.
[Jim] It is actually a map you pass
in. Our configurations are really key value pairs. It would be nice to add
or remove a configuration element.
If the requirement was you needed to reconfigure the entire
settings, that might be horrible if you have multiple threads using the
component or have requests in process and reconfigure. Some of these maps
contain instances of objects...
[Jim] This is Dale. There
are two other API's I worked with that de-referenced
things. In both cases, in the directory system we went the route to have
it be a configuration setting, where as in the file system, it is a parameter in
the API. In this particular case,
management utilities need to be able to do both quickly. Ultimately need to
provide both for management utilities. They want to see and then
manage.
[Drummond] It doesn’t seem like you
can do this entirely through settings.
[Jim] The settings thing is great,
but need to set option, do function, then set option back. That runs into
3 API calls.
[Dale] But it also means you can
have so much magic context that it can be hard to debug.
[David] There are some things you
don't want to page. At some point will need Context Providers that can do a
page. Can’t do this thought
configuration.
[Jim] Something that complex will by
definition require new API’s. If we were able to do that
with just configuration, it would look pretty
hole-ey.
[David] In the past, pass
in key value pair object that is API
specific.
[Jim] First argument is to add more
arguments to the method. It can explode out. It solves the issues that
Dale brings up. You are just calling
one API and in terms of debugging, it
is not some magical state tweak prior to the API being called. So I'm
hearing that using the more dynamic setting isn’t the best way to approach
this.
[Dale] Dynamic settings eventually,
but not solve this problem with dynamic settings..
[Jim] My biggest issue with adding
one more argument is that there are 100 methods.
[Paul] Why do you need to add it to
all, why not add it as needed?
[Jim] If there is anything in that
extension argument, can just say I don't support anything in that argument. That would give us some practical
experience.
[David] The Context Provider could
set things in those arguments as they turn off as well. The scenario I keep
thinking about is paging....
[Jim] I put some stuff back out on
the list if people are willing to look at it. I know you have helped a lot
in the past David Lets all go in that direction, and make sure I get your input
before I check-in anything. That is all I had on that one. That gives me enough
to do for a few days..
[Mary] Are there other
topics?
[Paul] I have a question
for Jim. There was a bugzilla
item to add a reference type attribute. Can you confirm that what we are
talking about is 3 things: literal, object reference, and a new kind, a defined
kind of literal known to be a reference.
Did I capture the intent?
[Jim] Yes. It may be a literal or a
subtype of a literal. I don't know if you looked at the Higgins HOWL stuff.
There are three literal types that are referenced. Have a look at those. Maybe they will satisfy what you want. I
assumed you would just close that bug.
But thought maybe you were waiting for a comment on the entityUDI page,
the canonical format.
[Paul] On the Higgins entity page
wiki.
[Paul The reason I didn't close it
is because I hadn't heard back from you. I think I did what is
needed.
[Jim] I Like that doc.
[Paul] I will send that
clarification to the list.
[Jim] If I was looking at my Jim
entity with a home address attribute pointing to another entity representing my
house, what data type would that be? Would the data type be
entityUDI?
[Paul] It depends. If you had the
setting on to resolve to an actual iEntity, then return value is. else an
entityUDI. I wasn’t sure how you were gong to model that.
[Paul] I will fix the bugs in the
doc.
[Jim] It will be the official page
for the definition
[Paul] When have the
API change done it is a major step
forward for IdAS.
[Jim] I want to think more today
about what one of these extension things looks like, get feedback from the list,
then find a CP to implement it in.
-end 12:59.
Links
=====
[1]
https://bugs.eclipse.org/bugs/buglist.cgi?query_format=advanced&short_desc_t
ype=allwordssubstr&short_desc=&classification=Technology&product=Higgins&tar
get_milestone=1.1M2&long_desc_type=allwordssubstr&long_desc=&bug_file_loc_ty
pe=allwordssubstr&bug_file_loc=&status_whiteboard_type=allwordssubstr&status
_whiteboard=&keywords_type=allwords&keywords=&emailtype1=substring&email1=&e
mailtype2=substring&email2=&bugidtype=include&bug_id=&votes=&chfieldfrom=&ch
fieldto=Now&chfieldvalue=&cmdtype=doit&order=Reuse+same+sort+as+last+time&fi
eld0-0-0=noop&type0-0-0=noop&value0-0-0=