Discuss the 3.2.4 release

[frank]

Discussions about the maintenance release can be added to this topic.

[Spyrus]

Nice, but a bit early for this? Only 5 days after previous bug release? Is there any serious bug fixed we are not aware of?

[James Bond 007]

Yeah, it seems to me that VirtualBox 3.2 was not released that long ago, (on 18 May, less than a month ago) and already 2 maintenance releases?

Seems that the 3.2 series is really buggy. Does Oracle/Sun not have enough testers?

More seriously, this is not a good sign. A major update followed quickly by not just one, but TWO maintenance releases in quick succession. The quality of the 3.2 version does not give confidence to users at this time, do you agree?

[takwing]

Franky speaking, v3.2.4 is worst than v3.2.2 in the support for Mac OS guest.
Other than this, I am not aware of any big issues in upgrading to v3.2.4 for Windows guest so far.

I am not within the VB team but I believe the VB team has been trying their best to get a balance between the amount of features/fixes add to each release, regression testing and the release schedule.

Virutalization is really a hard topic, I always believe that it depends heavily on the actual hardware.
VB3.2 is a big leap.... and having 2 maintenance release in a short period of time is probably due to the face that the team "hears" the noise from the users and response quickly....

My only comment would be "can we have beta release also for the maintenance release?".

[takwing]

Thanks the team for promptly response to the problem of "booting up Mac OS X guest via EFI" by providing us with build 62467!

I am not sure if there is anything else fixed in this regression.... what I can confirm is that the EFI problem is fixed.

[frank]

My only comment would be "can we have beta release also for the maintenance release?".

All this is right and we normally would have done a Beta release (and sometimes we did). Unfortunately the OVF bug in 3.2.2 was quite nasty and required a quick response. The result was the broken EFI in the 3.2.4 build

[Perryg]

@Frank,

Does this (June 8th) build have anything else in it besides the EFI fix? Is there any problem running the (June 7th) build if you do not need to deal with EFI, or is it suggested to just update to be sure?

[frank]

Actually there is a small fix for this issue as well (X11 hosts only). Users which are not affected by neither the EFI problem nor by this Alt-TAB problem are fine with the 62431 build.

[Perryg]

Ah ha! Thank you my friend. I shall install the new build and continue on with my tests. One last question.
The new (June 8th) builds do not trigger a reinstall of the GAs. Is it safe to leave these alone or should they be forced to reinstall?

[qkb]

The text on the main download page is a bit confusing:

"Note: The original 3.2.4 packages (build number 62467) contained a bug in the EFI BIOS. The build number of the fixed packages is 62467. Sorry for the inconveniences!"

Is 62467 the buggy built or the fixed built? Or both?

[Snolen]

62467 is the fixed build (62431 was the buggy).

http://forums.virtualbox.org/viewtopic.php?f=1&t=31799

Regards.

[qkb]

62467 is the fixed build (62431 was the buggy).

http://forums.virtualbox.org/viewtopic.php?f=1&t=31799

Regards.

Yes, I know, I've noticed the buggy number on the forum, but if someone just jumps to http://www.virtualbox.org/wiki/Downloads (from main page to 'Downloads') then s/he is greeted with the confusing message.

[frank]

The text on the main download page is a bit confusing ... Is 62467 the buggy built or the fixed built? Or both?

Fixed. Thanks + sorry.

Perry: Right, the were really no changes to the guest additions .iso image.

[frank]

See this report (Alexander is one of our developers). This bug is relevant for all users of an ext4 file system on the host and if the host cache is disabled (default for SATA / SCSI)!

Therefore our recommendation: Enable the host I/O-cache for all your VMs if your host file system is ext4! I believe, mainly Ubuntu 10.04 users are affected.

[Technologov]

Indeed, yes, I feel that v3.2.x line has quite many regressions, despite I done work during the BETA phase.
we had so far (after BETA), v3.2.0, 3.2.2a, 3.2.2b (with Windows bridge fix), 3.2.4a, 3.2.4b (with EFI fix)

May I really *ask* _please please_ don't introduce regressions in maintenance releases.
Among other things, it means please don't change code if not absolutely necessary.

For example: Can you please change EFI only for next major release skipping changes in this v3.2.x line ?

Unless critical or blocker bug is found, (such as the OVF one), or some very trivial bug, that has no chance for regressions, such as text bugs, most medium bugs could be pushed to next major release.
Would such a conservative policy (of non-regressions) be a good idea for this project ?

About the ext4 bug - hugh - I have to say "no words, only letters". Sad.
Really. What can VBox project do about it?
1. revert default to always enable host I/O cache ?
-or-
2. VBox Linux installer could check for ext4 filesystems on host.

-Technologov