[Cosmo-dev] [Sum] Nov 26-Dec 2

[Ted Leung]

Here's the summary of the activity on cosmo-dev for the week of   
November 26, 2006.

New threads:
-------

Brian Moseley posted his EIM status to the wiki: <http:// 
wiki.osafoundation.org/bin/view/Journal/BrianMoseleyCosmoEIM>
+ Morgen Sagen responded with an updated Chandler page: <http:// 
wiki.osafoundation.org/bin/view/Projects/SharingProject> and a  
question about only sending modified records instead of all EIM records
+ Brian replied that he hadn't planned to handle only modified  
records for Cosmo 0.6
+ Morgen agreed to sending all the records
+ Brian also posted about a RELAX NG schema for eimml, and moved the  
EIMML spec to <http://wiki.osafoundation.org/bin/view/Projects/ 
EimmlSpec>
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-November/ 
002130.html>

Randy Letness committed a bunch of changes to the stamping branch
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-November/ 
002139.html>

Some leftover JSP Tags from tiles were found in the code
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-November/ 
002146.html>

Philippe Bossut pointed out DateBocks, a library for natural language  
date parsing
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-November/ 
002147.html>

Travis Vachon and Matthew Eernisse surfaced their discussion of the  
dynamic CSS styler:
+ There was some discussion of the best way to fix some ugliness with  
the styler
+ There was also some discussion on a better way to implement app  
wide constants
+ A number of libraries will get Dojo-ized, except where the  
equivalent Dojo library is buggy.
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-November/ 
002150.html>

Travis Vachon and Matthew Eernisse discussed the dueling stylesheet  
problem, left over from the Cosmo/Scooby merge
+ a Dojo boilerplate JSP tag is part of the solution
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-November/ 
002156.html>

Bobby Rullo raised concerns about the 0.6 implementation of stamping:
+ His main issue was that each stamp type is a separate class and  
database table.  He was concerned about storing previously unseen  
stamp types
+ Brian Moseley responded that this model was taken from Chandler's  
model
+ Randy Letness pointed out that Cosmo can persist arbitrary data  
using item attributes, but that Cosmo also did not have a way to  
persist arbitrary item types.  He pointed out that there is a  
tradeoff between data model flexibility and performance, especially  
in a multi user system
+ There was also some discussion about the desirability of a parcel  
like extension mechanism for Cosmo
+ Ted Leung asked how the implementation would handle things like JPGs
+ Randy Letness said that Cosmo can persist JPG's sent via dav, and  
that new stamp types might not require a new table if the new stamp  
type stores stamp data as item attributes, which is what Chandler does.
+ Phillip Eby pointed out that EIM has a data type for blobs
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-November/ 
002165.html>

Ted Leung and Matthew Eernisse discussed adding bug 6972 to the Cosmo  
0.6 feature list (it will be added):
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-November/ 
002177.html>

Matthew Eernisee forwarded a message indicating that  
dojo.html.common, not dojo.dom was the correct package for Dojo's DOM  
manipulation library, at least for HTML
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-November/ 
002183.html>

Brian Moseley started a thread about simplifying eimml:
+ The crux of his argument is that EIM is actually more expressive  
that what Cosmo need to do in the short to medium term, and that the  
sharing work use several fixed data models for the time being.
+ Philip Eby pointed out that a lot of the algorithms around EIM/ 
EIMML would have to be implemented for each of the fixed data  
models.  Also Chandler will use EIM for dump/reload.
+ Brian Moseley didn't know that Chandler was going to use EIM for  
dump and reload
+ Philip Eby posted more details on the EIM centric syncing  
algorithms that he and Morgen Sagen have been working on.
+ There was also discussion of grouping EIM records related to a  
particular item as well as metadata EIM records.
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-November/ 
002182.html>

Bear (Mike Taylor) posted a status report on the Maven 2 conversion:
+ He's a good part of the way there and needs some help from the  
engineering folks
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-November/ 
002193.html>

Travis Vachon added a new property to cosmo.properties and a new  
query to the data model.
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-December/ 
002196.html>

Aparna Kadakia started a discussion on how we plan to test Cosmo's  
data migration features.
+ Randy Letness responded with some ideas
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-December/ 
002209.html>

Katie Parlante described the Chandler team's plans with regard to  
implementing the new sharing format.
+ One area of concern was around changing EIM (domain model) records
+ Brian Moseley indicated that changing EIM records was very low  
impact but that changing EIM was high impact.
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-December/ 
002210.html>

Bear (Mike Taylor) called for a vote to switch to Maven 2:
+ Brian Moseley preferred to wait until Maven 2 can clean the in- 
place war.
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-December/ 
002211.html>

Ted Leung posted the notes from the November 30 Cosmo Engineering  
Meeting
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-December/ 
002215.html>

Matthew Eernisse committed the cosmo.app module:
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-December/ 
002218.html>

Matthew Eernisse posted some tips for using programmatically created  
widgets so that the whole page is not re-flowed:
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-December/ 
002219.html>

Continuing Conversations:
-------

The thread on benchmarking Dojo event/widget performance continued:
+ Matthew Eernisse agreed that Dojo's addition of lots of properties  
to a widget was not a performance problem.  He still felt that it was  
inefficient for a developer
+ There was some further discussion on having some performance tests  
to track these kinds of issues
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-November/ 
002131.html>

Ted Leung created a bunch of new bugzilla components:
<http://lists.osafoundation.org/pipermail/cosmo-dev/2006-November/ 
002138.html>

Announcements:
-------