Uwe,
All in all, there's seems to be quite many people interested in 6 month
cycle. Well, I'm fine with that, it's just that this is not a simple
decision. Not using 3 months has some disadvantages as well.
On 12/18/06, Uwe Altmann <[EMAIL PROTECTED]> wrote:
OK; I see your point. But you just gave the second best answer.
The point so far is: QA (at least Mac QA) is not longer able to test all
these funny new things all these engaged programmers write.
So we can not keep up the pace the project goes.
Well there are several levels of QA. We may not have time to do the Full QA
for every release, but we currently do much QA for each release. I doubt
there will be this much activity in the middle of OOo 2.3 development (the
half-way of 6 months). This means the bugs will hit us 3 months later, and
there will be twice many of them (2* three months).
So, what we will get by your proposal is a three class "society": First
class: Windows & Linux - QAing every release, making the pace; 2nd
class will be Mac & ??, leaving out every second release and 3rd class
would be more seldom OSses with just one release every year or so.
My proposal is done from a practical view. I do not want to impose any
classes or hierarchy. Everybody contributes as much resources as they have.
Obviously Sun-paid engineers have lots of resources. Having QA'd Windows and
Linux builds in several languages is already a great win for Mac OS X. Lot's
of bugs get fixed from that QA.
But the better way would be (at least imho) that we think about another
approach: *Less coding - more QA* . Reads: Putting less resources in
coding and more resources in QA.
OK, QA is not such a funny thing as coding. But it's surely the better
way to a high quality product.
Perhaps the project should even do more reengineering (I know, that's
an even worse job than doing QA). I remember a mail on [EMAIL PROTECTED]
"Thoughts about a world without VCL"
(http://porting.openoffice.org/servlets/ReadMsg?listName=mac&msgNo=3277).
Very good argumentation by Christian Lippka.
I have no interest in doing "perfect" QA for a product that has very bad
user experience. As long as OOo for Mac is not native and polished, it
continues to look "Alpha" or "Beta" version in my eyes. And betas do not
need "perfect" QA.
I have already commented on Christian's ideas at that time. And I am the
only one who has worked on porting the ooo (cairo) canvas to Mac OS X. So I
am doing (a small part in) the re-engineering.
However, the more I look at the (sometimes ancient) code that is the GUI of
OOo, the more I have the feeling, we should just use OOo as a library and
then build the UI from scratch using XCode et al. (This is the approach
Mozilla's Camino is using). I seriously doubt there is any chance of getting
OOo designed as a proper Mac app, using the existing OOo technology. Unless,
that is, we want the OOo Mac to look exactly like Windows OOo, except for
some Aquafiled buttons.
If OOo has been so excited at copying and mimicking the UI of MS Office, why
not then go all the way and mimick the Vista Office for Vista (the Ribbon
UI) and iWork for Mac OS X. Even that (while quite herculean engineering
task) would be better than the current situation.
Me personally can live happily until the end of my days with "slowing
down the development" as you wrote.
I agree so far that the pace of writing new code is and must be the pace
of the whole project - but my answer is to slow down exactly this and
use the then free resources on other places where they are needed
urgently.
Linux kernel did not get popular and widely used, by slowing down its
development. Neither did it help Mozilla to have milestones every 1-2 years
(the time before Firefox and Phoenix got the main attention).
While everybody should do QA at some level (most engineers QA the code
quality), there is no point in having armies of QA people. If there is a
problem with the current QA situation, it is the fact that the QA process
itself is too heavy.
We need better QA tools. One idea would be automatic QA for nightly builds
(or in OOo, the milestones). Also for Mac OS X, not just the Sun blessed
platforms.
Anyways,
Time to go back to doing some real stuff.
Mox