Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
[hyades-dev] Meeting minutes for October 28th Hyades Execution Team Meeting, including Draft Datapool API for Review

I have attached the meeting notes for this week's execution team meeting.  I have also attached the first draft of the datapool APIs (courtesy of Joe Toomey and Bill Tobin) we briefly reviewed during the meeting, and for convenience also the associated draft datapool feature design doc that Joe Toomey sent out a few weeks ago to hyades-dev.

Please review the datapool API HTML documentation (in the zip file), and direct your comments back to Joe Toomey or to hyades-dev.

I believe Joe plans on holding another review of this during the weekly Test Model meeting this Thursday, with a different emphasis.

Thanks,
Kent 


Execution Team Group Weekly Meeting, 11AM Eastern, Tuesday, October 28, 2003

Attending
---------
Kent Siefkes, IBM Rational Raleigh
Joe Toomey, IBM Rational Lexington
Javier Garcia, IBM Rational Raleigh
Antony Miguel, Scapatech
Stephane Leroy, IBM Rational Toulouse
Andrew Hoppe, IBM Rational Raleigh
Todd Merriweather, IBM Rational Raleigh


Proposed Agenda
---------------
1. Review 1.2 execution schedule, including feature-based (cross-group) schedule for datapools (Joe) and HTTP (Todd)
2. datapool runtime API review (Joe)
3. Revisit status/recap of Antony's suggested execution API enhancements


Recorded Discussion/Decisons (new discussion/status preceded by *)
------------------------------------------------------------------
1. Review 1.2 execution schedule, including feature-based (cross-group) schedule for datapools (Joe) and HTTP (Todd)

	Execution deliveries for 1.2 that have committed staffing:

	- Document 1.1 ability to execute tests standalone (command line) - Joe
		* Move from Iteration 1 to 2

	- T4 (P2 1.1) Support a more interactive control model that allows execution to not only be started,
	  but also paused, restarted, stopped and cancelled. This will include events to reflect these
	  execution state changes. - Joe, Antony (design only)
		* Coding not started, targeted for Iteration 2

	- Datapool runtime API (subset of T5) - Joe
		* On track for design and coding
		* Datapool UI Editor staffing still uncommitted

	- T3 HTTP load testing (beta P1 1.1, full function P1 1.2) - Todd project manages, multiple contributors
		HTTP enhancements (from http plans submitted to hyades-dev on Aug. 26th):
		- Cookie Handling
			* Iteration 1, on track for 11/7 code completion
		- SSL playback
			* Iteration 2
		  - OK to require JRE 1.4 for SSL record/playback, but must record/playback non-SSL with 1.3
		  - design with idea of leaving open the ability to provide a pluggable security provider later,
		    either use of RAC's (undocumented) model for pluggable security provider orsomething more
		    limited
		  * - change to keep persistent connections across pages (also applies to non-SSL)
		- User Timing Control (think times) [already submitted in 1.1]
		- Data variation (datapooling)
			* Iteration 2 [no extra work beyond T5, covered above]
		  - OK to require manual Java coding (using datapool runtime APIs) to achieve this
		- Page level response time measurements
			* Iteration 1, on track for 11/7 code completion
		- Simple scheduling (workload definition that uses individual http tests as building blocks)
		   - OK to use existing ability of a Hyades test suite to call other test suites, so this would
		     be satisfied by having an HTTP test suite call other HTTP test suites
		   * Just needs testing to verify how well this works
	
		* Brought up issue related to heirarchial event support for HTTP multi-user test runs - Iteration 2.
		  Todd will follow-up.
			(runner - Ashish, loader - Ashish, Event History UI - Brian Battersby)



2. datapool runtime API review (Joe)

		*See associated HTML doc. We concentrated on the runtime interfaces, although interfaces for creating
		and editing were also included.  There was not alot of detail feedback because this was the first
		exposure to the API.  One change request was raised by Kent during the meeting: the IDatapoolRecord
		interface did not have a method to retrieve the value of a cell within the record, indexed by name (only
		one that indexed by position - getCell).  The IDatapoolIterator interface did have such a method, and
		Joe thought it was an oversight that IDatapoolRecord didn't.

		*There was also some feature discussion regarding import/export to other persistence formats, such as .csv,
		but this was a separate topic and had no impact on the test model or API design.


3. Status/Recap: Antony's suggested execution API enhancements

	Antony (or scapatech) will contribute to interface specification, may (in 1.2 or later) contribute to
		implementation if needed for a later contribution and it hasn't been done yet by anyone else
	Priority: After T7
	No one has signed up for contributing the code to go along with the interface specification yet, and there
	remains concern about providing the spec with no default implementation provided, so this may not go into 1.2.
		* No staffing yet for this
	
	Based on design discussion at 10/7 meeting, Antony updated the proposed API on an email to hyades-team on 10/8
	(add boolean, define class with 5 data types vs.using object so that existing marshalling can be utilized).

	For the execution history part of Antony's email, that belongs in the test model discussion and in
	Richard Duggan's execution history event sub-group.
		* Antony and Richard have discussed, but not committed yet (relates to perfmon contribution)


 
New Action Items
----------------
* None

Attachment: Datapools.zip
Description: Zip archive

Attachment: Hyades Datapool Feature Design.doc
Description: Binary data

Back to the top