MozillaZine

Drivers Update

Friday March 29th, 2002

The mozilla.org Status Reports was updated, and included some information on what drivers@mozilla.org are up to, and what's coming up:

"drivers@mozilla.org approved checkins for over 240 bugs this last week. We've taken a few more big-ticket items and have a couple to go. drivers will be ramping down this next week and actively soliciting fixes that we need to branch and have a Release Candidate 1 sometime (hopefully) late next week. We need to get RC1 out there quickly and get feedback on all the changes we've taken since 0.9.9 and at the same time we need the release to stand up so that users will put enough hours on it to give meaningful feedback. Please help us to get the builds into good shape over this next week with particular attention to recent regressions and topcrash+ bugs."


#30 Do not release binaries of 1.0 code

by ipottinger

Monday April 1st, 2002 12:31 AM

You are replying to this message

*** RELEASING BINARIES GENERATED FROM THE 1.0 CODE IS POINTLESS SINCE, BY DEFINITION, THE 1.0 CODE REQUIRES NO FURTHER TESTING.

This is assuming that prior to the release of the 1.0 code, all known and newly reported bugs would have been deemed to be either: 1) resolved by a (trivial) fix that requires no further testing of the code, or 2) not a \"show stoppers\" and hence will be tolerated within the 1.0 code. If any bug had not met one of these two conditions then, assumingly, development would have continued after which binaries of a new release candidate code would have been offered for testing and reevaluation of all new and know bugs would have begun again.

Releasing 1.0 code binaries might allow previously unknown bugs to be discovered, but to what end? The 1.0 code, by definition, is not a release candidate to be altered and retested. \"1.0\" is, however, a point in the development of code when testing is stopped and, even though known and unknown bugs might still exists, the code is released \"as is.\" With a new round of contributions, hacks and bug fixes, development of the 2.0 code will begin. Then and only then, in the 2.0 development cycle, should new binaries be released and testing begun anew.

*** MOREOVER, BINARIES OF THE 1.0 CODE ARE COUNTERPRODUCTIVE TO MOZILLA EVANGELISM.

Since, as stated above, 1.0 binaries can play no useful role in the development of the 1.0 code, their existence can only be a source of confusion that promotes the misconception that mozilla is a end-user product instead of a code base on which a end-user product can be built. This misconception leads many to evaluate the mozilla project in an inappropriate fashion, often resulting in bad press or negative word of mouth that paints the entire project in an unfavorable light. Already, too many individuals look upon mozilla.org as the producers of a browser suite especially since the publicly available test binaries (milestones) make it so obviously clear that the code is being developed and tested in that form. These same individuals go on to complain that \"the mozilla browser\" is incomplete, missing key feature found in most browser suites like spell checking, encrypted email, etc. With that in mind, a release of 1.0 code binaries could be the source of more bad press that, though unfortunate, will be understandable and, more importantly, *avoidable*.

Take, for example, the developer of the spell checker code licensed by AOL/Time Warner and combined with mozilla code to form the feature rich Netscape 6.x. If besides spell checking, this company also produces code for page-layout, printing and other word processing feature (I\'m not sure, I don\'t know who they are) then developing and testing all that code in the form of a word processor might be the must efficient and effective way to proceed. Providing their code and test builds to the public, as mozilla.org does, might bring the benieft of thousands of extra hands and eyes poking and scanning for errors along with the odd complaint from a confused few about missing features in \"the company\'s word processor.\" However, if that company announce a point-oh release of their code and then provided word-processor-like binaries of it, then they would be actively courting disaster. Since the binaries could not possible be \"test builds,\" the only possible purpose the public could assign them is the representation of the company\'s effort to produce an end-user product. The avoidable unfortunate feedback, regardless of how harsh, would be understandable.

Hopefully, mozilla.org soon will also be the producer of spell checking and email encryption code. The producers of products like Netscape 6.x will need not build or license from others as much code as they do now and mozilla.org will be increasing seen as a one-stop open source of useful code. WE CAN ENCOURAGE THAT PERCEPTION IF THE CODE AND ONLY THE CODE IS PRESENTED TO THE PUBLIC WHEN IT IS DECIDED THAT 1.0 HAS BEEN REACHED.

----- I\'ve submitted the above is the text as bug 134597 since I want to know mozilla.org\'s opinion on my arguments. Go there to vote if you agree but please post your public comments in this forum or email your private comments to <ian@pottinge.ca>. Thanks