Mitchell Baker on mozilla.org and Phoenix
Saturday October 5th, 2002
Mitchell Baker, mozilla.org's Chief Lizard Wrangler, has posted a newsgroup message about mozilla.org and the Phoenix project. The post explains why mozilla.org is supporting Phoenix and how the development will affect the main SeaMonkey project.
I do agree with the posting, but a few things are not clear at all: 1) Will Phoenix be "ported back" to Mozilla (ie, integrated)? 2) Will Minotaur (Mail/News equivalent to Phoenix) be integrated in Mozilla sometime in the future? 3) What about Composer (with all its links to Mail/News codebase)? 4) What will happen to all the existing extensions (Enigmail, Cascades, etc.)?
Seems to me there is a lot of duplicated effort: developers fix bugs in the UI for the trunk, while other engineers are developing a complete substitute. Don't get me wrong, I really like Phoenix and I have a few old machines which will benefit greatly from a standalone browser, but I fear that having more and more branches in the main Mozilla codebase will end up in a very-difficult-to-manage situation. Look at NS7: in many tests it outperforms Mozilla (even latest builds) and the relative fixes didn't find their way to the trunk. No flame involved, just some cinic thoughts.
There is a very small number of engineers involved in the Phoenix project. A number of them are in university or are employed by companies other than Netscape. I don't see any problem with these folks working on different solutions. Phoenix is built off the daily Mozilla codebase (from what I recall), so Phoenix will at least be keeping up with the changes in Mozilla's basic functionality.
I think that innovations/changes that occur in Phoenix would probably make their way into the Mozilla codebase on a case-by-base basis. Either that, or the Mozilla team will evaluate the Mozilla/Phoenix situation at a later time and decide to make a wholesale change to system revolving around standalone components. I think there are a lot of positive aspects to the Phoenix model that's developing. The Mozilla engineers seem to be a practical lot, and I doubt that they would discriminate against changes that were beneficial simply because they first found a home in a different branch of the Mozilla codebase.
#3 I'm all for standalone apps
by _rgw_ <firstname.lastname@example.org>
Sunday October 6th, 2002 9:20 AM
Phoenix is rapidly becoming faster, smaller, and more innovative then main Mozilla trunk. If Minotaur can produce the same results then I think every component should get some standalone lovin'. Personally, I would like to see Composer as a standalone. It would be kind of like Edit Plus 2.
#5 Others may be, too...
by choess <email@example.com>
Sunday October 6th, 2002 3:15 PM
When I went to the East Coast Developer Day last spring, people were very keen on the idea of "disentangling" the browser suite into proper, potentially standalone apps. (This was part of a general discussion on what things should live in the mozilla CVS tree, the mozdev CVS tree, etc., and what should build-by-default.) I wouldn't be at all surprised to see Mozilla-the-browser-suite become a bundle of standalone apps.
#6 De-integrating Mozilla components
Sunday October 6th, 2002 5:22 PM
This would be a smart move on the part of the Mozilla team. Small building blocks from which you can chose the ones you need have always been the way to go in software as opposed to a monolithic piece of code ("bloatware" as it is so often put). I wonder how easy it will be to separate the components though?
#7 Integrated doesn't mean bloatware
Sunday October 6th, 2002 7:12 PM
You can already install the Mozilla components you need. On the opposite running some of these standalone mozilla apps at the same time will be much more resources hungry than just running an integrated suite.
As a Netscape employee I'd love pointers to any tests where N7 comes out better than Mozilla. Hard to believe it could be true, though, since both are built on the same build machines at the same time from the same source. If anything you'd expect the added Netscape components and overlays to have a slight negative impact on performance.
Cnet.com has a benchmark that put Netscape 6.2 faster than Mozilla 1.0 in a couple of tests that itself deemed as margins too small to be conconclusive.
The test results: <http://www.cnet.com/softw….dir.3227884-8-20005816-6>
I'm sorry for answering this late, but my machine died and I'm still trying to resurrect it. I wrote about NS7/Moz speed differences after reading a bug (sorry after so much time I can't remember the #) regarding DHTML (search for perf related bugs fixed or partially checked in in the 2 days before my last post), and a comment had some vastly different results (IE, NS7, Moz 1.1 and trunk were compared). As soon as I can get back a really working machine, I'll add more details.
#10 Re: Fine, but...
by mitchell <firstname.lastname@example.org>
Monday October 7th, 2002 4:59 PM
We can't really say yet what the future will bring. What will happen with Minotaur? We don't know yet, we'll have to see how much attention it gets and what the results are before we can make long term plans. Same with Composer. To date, I'm not aware of an analagous effort to make it stand-alone and small. Right now we want to see what happens with Phoenix, and hopefully Minotaur. When there is more data we can do more planning.
Phoenix is not in a seperate branch, and there is little duplicated effort going on. it lives in $src/mozilla/browser. it uses features that mozilla-proper uses plus some of its own features. mozilla-proper is free to use any feature in $src/mozilla/browser as soon as it wants to, and phoenix gets mozilla-proper features as soon as they land.