Pamela J. Taylor, SHARE
Roy Reed, Sun Microsystems
Michael Iarrobino, Sun Microsystems
Chris Dollin, HP
Russ Daniels, HP
Waldemar Horwat, Netscape
Patrick Beard, Netscape
Markku Vartiainen, Phone.com
Dave Raggett, HP
Jim Miller, Microsoft
Joel Marcey, Intel
Carol Thompson, HP
Mike Deese, Microsoft
Sam Ruby, IBM
Mike Cowlishaw, IBM
Rex Jaeschke, Microsoft
Herman Venter, Microsoft
Jeff Dyer, Compiler Co
Andrew Clinick, Microsoft
Clayton Lewis, Netscape
Jan van den Beld, ECMA
Louis Theran, Nokia
Bill Gibbons, Pixo
Everyone does a quick introduction to themselves.
Document covering the Microsoft proposal.
Andrew volunteers to write up and circulate the minutes himself.
None.
Sam Ruby reports that meetings have been lively. The target is to approve the spec at the March 2001 TC39 meeting.
Markku Vartiainen reports on the proposal for a Mobile Profile for ECMAScript. Two possible avenues: a technical report on its own or as an appendix of ECMA 262.
ECMA 262 is currently on a fast track to become an ISO standard. A technical report is not a mandatory report.
Bill suggests we might first generate a technical report and then work on making it into a normative annex to ECMA 262.
The agreed plan then is to produce a technical report for a subset of E3 and to migrate this into a normative annex for E4.
Action: Andrew Clinick to email the TC39 list with the announcement of the formation of a subcommittee tasked with the development of a technical report for the Mobile Profile. Dave Raggett volunteers to lead the subcommittee.
We then turned to discussion of the comments on E3 received from Japan. Andrew volunteers to own the process for dealing with these comments.
Co-sponsors include HP, Intel, Microsoft and Fujitsu.
Copies of the slides are available on the ECMA FTP site.
Mike Cowlishaw asked what the task groups would be doing after the final drafts are submitted to
Bill Gibbons: conventional wisdom has it that standardization shouldn't start until there at least 2 independent implementations. What is the situation with regards to C# and CLI.
Tony: Microsoft is working on two implementations.
Bill: it would be good to see an implementation from another company such as HP.
Tony: we are interested in working with other companies to encourage the development of other implementations.
Jim Miller: Microsoft's two implementations are largely independent, although there is some code overlap for the class libraries.
Tony: we are working towards an open source implementation, but we don't intend to tag this as a "reference implementation".
Patrick Beard: can you comment on the man-years of effort involved in developing the compiler and run-time?
Answer: 10 to 12 man-years for writing the specification and doing the compiler implementation. We estimate it would take roughly 2 to 4 man years if you start from the specs.
Jim Miller: the majority of the effort is in the class libraries. The CLI involves a class loader, a garbage collector and optionally a just in time compiler. The JIT alone is about 2 man-years for people with previous experience.
Pamela Taylor: why should this work be done in TC39?
Tony: we see a strong overlap between the goals for CLI and the current work on ECMAScript.
Waldemar Horwat: CLI doesn't appear to be intended to support dynamic languages, and sees this as causing problems for ECMAScript.
Andrew Clinick: Microsoft is willing to work on making CLI better suited to dynamic languages. He recognizes the importance of dynamic aspects of ECMAScript.
Waldemar: can you provide assurances that ECMAScript E4 will be dynamic?
Andrew: our customers are asking for the benefits of being able to statically compile ECMAScript.
Jim: Microsoft is indeed interested in ensuring that CLI is appropriate for dynamic languages.
Chris Dollin: what can we do to ensure that people working on CLI are in touch with work on E4?
Jim: we would like to have the technical meetings on successive days so that people can attend both meetings.
Clayton Lewis: Netscape hopes that CLI succeeds and plans to participate. C# and CLI are not scripting languages, and would the work not to take place in TC39 which is chartered to work on scripting languages.
We have repeatedly seen problems over the last year in attempts to progress the dynamic aspects of the E4 proposals for ECMAScript. We see an impedance match between ECMAScript and C#/CLI.
Netscape wants work to start on C# and CLI in TC39, but once the work is underway, Netscape would like the work to move to a new TC in order to minimize the tensions with work on ECMAScript.
Pamela: is CLI available on non-WinTel platforms?
Tony: Yes, but Microsoft can't comment on this at this time. Toby would love to see other companies involved.
Waldemar: are all CLI implementations expected to include a large number (600) of standards?
This would be up to the committee to decide. A layer cake approach may be appropriate.
Patrick: can Jim comment on the match of CLI to dynamic languages?
Jim: there are a few corners to round off, but these are in the class libraries. We would like to deal with these in the 2nd edition of the specs.
Mike Cowlishaw: is there any intent to align the data types with say XML Schema?
Jim: That sounds like a valuable idea.
Jan van den Beld: 4 sponsoring companies is sufficient for work to start in ECMA, but it would be interesting to hear from other companies.
Sam Ruby: IBM would like to participate.
IBM, Netscape and Sun Microsystems would like to participate.
Pamela: SHARE is interested
Bill Gibbons: Pixo is likely to be interested.
Microsoft circulates the following text:
TC39 agrees to undertake into is Programme of Work the development of standards for C# and CLI as proposed in the document TC39/2000/2 and further clarified by TC39/2000/4. Furthermore, and recognizing that the undertaking of this new work will likely result in substantial additional resources being added to the Committee, TC39 anticipates the need to create TGs within its structure to facilitate both rapid deployment of the new work and the continued emphasis on its current work. TC39 resolves to implement such a new structure in a manner as to cause minimal impact to its organization and ongoing Work Items. Finally, TC39 resolves to review both its structure and its term of reference and to prepare recommendations to the Secretary General respecting amendments if appropriate, by means of the TC39 chairman at the May 2000 GA meeting.
Jan: new TC's have to be approved by the General Assembly, but task groups can be set up directly by a TC. Let's us see how work proceeds, and we can then see whether a new TC makes sense. Perhaps we could move to create a new TC after Christmas if appropriate.
Rex Jeschke asks Jan to describe the ISO fast track process.
Jan: to be successful the fast track process relies on good work and avoiding surprising people. It may make sense to organize a workshop to involve anyone who is interested.
Roy Reed: Sun Microsystems supports ECMA taking C# and CLI on, but wants this to occur in a different TC from TC39.
Both Netscape and Sun are agreeable to work starting in TC39 provided there is a clear commitment to moving the work to a new TC in a reasonable time frame.
Mike Deese offers assurances on the proposed plan. He talks about the process for creating a new TC. He thinks that it would definitely take longer than six weeks to form a new TC.
Roy is still concerned over agenda item 6.3 and wants to revisit this after we cover agenda item 7.
... return to 6.3 ...
Mike Deese asked for TC39/2000/02 be registered as a document for this meeting.
Russ Daniels: this work may benefit from synergy between the task groups.
Mike Cowlishaw: but the task groups will be working in an atmosphere of uncertainty over their future in TC39.
Andrew Clinick offers assurances on avoiding interference with work on ECMAScript.
TG2: C#. - Carole Thompson (HP) as convenor and Rex Jaeschke (Microsoft) as editor.
TG3: CLI. - Sam Ruby (IBM) as convenor and Jim Miller (Microsoft) as editor.
Andrew Clinick says that this means we will be looking for a new editor for TG1. The convenor for TG1 will be Clayton Lewis, who takes over from Sam Ruby.
Carole and Jim introduce themselves. Sam and Clayton are already well known to TC39!
lunch break
???
Tuesday, March 13th in Redmond, hosted by Microsoft. The technical group meetings could follow on at the same location/week, and will be arranged by the convenors together with Andrew Clinick for Microsoft.
It is likely that TG2 and TG3 will meet in November since this is the month Microsoft plans to submit the C# and CLI specs to ECMA. The meetings will be in the week starting November 6th.