1.1.5 and 64-bit planning
Peter Hosey
reading the 7th version of the 64-bit patch.
Having just gotten past the halfway point last night, I changed my mind.
64-bit is not a 1.1.5 feature. We can wait until at least 1.2 on it.
Not too long, of course, because we don't know when Apple will release
Snow Leopard, but we need to worry about 1.1.5 first. (And then I will
get right back to reviewing the patch.)
So here's what I found on the bug tracker:
Bugs already fixed for 1.1.5:
- The GrowlMail crash
https://bugs.launchpad.net/growl/+bug/336354
- A crash in growlnotify, which someone reported on the forums and I
fixed in about ten minutes
https://bugs.launchpad.net/growl/+bug/336459
- GrowlMail posting notifications about read messages
https://bugs.launchpad.net/growl/+bug/251927
Bugs we *will* fix for 1.1.5:
- The off-by-two error
https://bugs.launchpad.net/growl/+bug/251376
(because it's so important)
- Prefpane still links to Trac
https://bugs.launchpad.net/growl/+bug/301955
All open bugs:
https://bugs.launchpad.net/growl/+bugs?search=Search
Feel free to nominate any bugs that you really, really want fixed in
1.1.5 and no later, especially if you're prepared to write a patch for
them.
Chris Forsythe
block it in case snow leopard comes out sooner.
Chris
Ned Holbrook
> 64-bit is not a 1.1.5 feature. We can wait until at least 1.2 on it.
> Not too long, of course, because we don't know when Apple will release
> Snow Leopard, but we need to worry about 1.1.5 first. (And then I will
> get right back to reviewing the patch.)
the possibility of a feature disparity between the 32- and 64-bit
versions of my application and the consequent costs to my employer in
terms of documentation and support should we beat Growl to the punch,
as it were. I certainly hope 64-bit doesn't get held up by GNTP.
Ned
Peter Hosey
> I certainly hope 64-bit doesn't get held up by GNTP.
It's not.
GNTP is important, as it resolves at least one Growl bug
(growlnotify's flakiness) and enables future cross-platform
interaction, but I have no idea when we'll finish it and am not
inclined to hold up a release for it.
64-bit is important, too, for app developers. And I would like to have
a 64-bit-compatible framework, even if Growl itself is not yet 64-bit
at the time.
But we have other things that are much more pressing for users today.
The big two: GrowlMail 1.1.4 crashes when not in Summary mode, and
users can't select the last two displays in the Applications tab's pop-
up menus. These are the things that outrank 64-bit right now.
As it is, people have been waiting nearly a month for the GrowlMail
fix, and Chris and I would like to keep each release small so we can
distribute it sooner.
Nick: If you can split that patch into two patches (one for the
framework+BeepHammer, one for the rest of the source), that would be
awesome. The framework patch would be much easier for everyone to
review and test, and may even be small enough to comfortably add to
1.1.5. We'll have to see it to make that determination. Sound good?
Chris Forsythe
To be honest, until 10.6 comes out, 64 bit is not a requirement
either. The day 64 bit is a requirement I intend on us shipping at
least the framework in 64 bit, if not sooner, however we have more
pressing issues (MAIL CRASHES, ZOMG) that are causing us headaches.
The faster we get those done, the faster 64 bit happens.
Our priority list is not clear at the moment, is my assumption as to
why you originally sent that email Nick. How can we address that?
Chris
Nick Zitzmann
On Mar 20, 2009, at 8:43 AM, Peter Hosey wrote:
> Nick: If you can split that patch into two patches (one for the
> framework+BeepHammer, one for the rest of the source), that would be
> awesome. The framework patch would be much easier for everyone to
> review and test, and may even be small enough to comfortably add to
> 1.1.5. We'll have to see it to make that determination. Sound good?
Since there's a considerable amount of overlap between the two, why
should we do this? If you want the NSInteger etc. stuff but don't want
64-bit yet, then we could always just turn off building for ppc64 and
x86_64 until you're ready to turn them on later.
Nick Zitzmann
<http://seiryu.home.comcast.net/>
Nick Zitzmann
On Mar 20, 2009, at 11:12 AM, Chris Forsythe wrote:
> To be honest, until 10.6 comes out, 64 bit is not a requirement
> either. The day 64 bit is a requirement I intend on us shipping at
> least the framework in 64 bit, if not sooner, however we have more
> pressing issues (MAIL CRASHES, ZOMG) that are causing us headaches.
> The faster we get those done, the faster 64 bit happens.
It's true that the 64-bit environment was immature and barely stable
back in 10.5.0 1.5 years ago, but a lot of the runtime bugs have been
fixed by now, and so I think it's RFPT. It isn't terribly critical for
GrowlHelperApp, yes, but we'd make a lot of developers happy if the
framework was ready sooner than later, since the framework's lack of
64-bit support is holding up some applications that can be ported over.
Nick Zitzmann
<http://seiryu.home.comcast.net/>
Peter Hosey
> To be honest, until 10.6 comes out, 64 bit is not a requirement
> either.
Not for GrowlHelperApp + extras, but the framework being 32-bit-only
prevents app developers from testing their apps on 64-bit. Waiting
until the release of 10.6 to enable them to test 64-bit will annoy a
lot of app developers, and some of them may simply drop Growl support
if we take that long.
There's a flip side to your statement that worries me more: If 10.6
will require 64-bit, and not support 32-bit, then we're in trouble if
Apple ships 10.6 before we ship a 64-bit-compatible Growl. A good
argument in favor of applying the whole patch at once, I suppose.
So I guess you've all convinced me: We should put 64-bit in 1.1.5. It
prolongs the current GrowlMail pain, but that pain is nothing compared
to the possibility of *no part of Growl working on 10.6*.