Closed Bug 145909 Opened 23 years ago Closed 23 years ago

browser.xpi corruption in Trunk and RC2 branch 0520 reported.

Categories

(SeaMonkey :: Build Config, defect)

PowerPC
Mac System 9.x
defect
Not set
critical

Tracking

(Not tracked)

VERIFIED WORKSFORME

People

(Reporter: tarahim, Assigned: jj.enser)

Details

(Keywords: relnote)

20020520 builds of both trunk and RC2 branch reports error upon installation. I have tried both full installer and stub installer, both failed. StuffitExpander reports a problem for browser.xpi upon expanding the full installer. Stub installer halts with an error number of four digits.(sorry I have lost it)
The builds I got from 20020520 worked fine. I ran both the 1.0.0 and the trunk installers with no errors. However, there were probably builds earlier than what I pulled ... I pulled the latest ones in a check made about 3 pm EST. The trunk build showed a 9am time stamp and the 1.0.0 build showed an 11:39 am time stamp in the "latest" and "latest-1.0.0" folders respectively. I have checked: the trunk build is 2002-05-20-03 the 1.0.0 build is 2002-05-20-05-1.0.0 Dale
My hunch is there is some incompatibility between latest Stuffit (6.5.1) and Stuffit Expander 6.0s. I am hoping if somebody responsible for the binary releases can check this out, because there are not Stuffit Expander 6.5.1 for every language yet.
Severity: blocker → critical
See also bug 148618
Cool! I've been having this problem for a while now, and upgrading to Stuffit Expander 6.5.1 (from version 6.0.1) fixes it. I saved the mozilla-mac-trunk-full.bin for build #2002062109. I attempted to unstuff it with Stuffit Expander 6.0.1. I got the following error message: "The integrity of the data fork of "browser.xpi" could not be verified. Use this file with caution." I then ran the unstuffed installer. It started installing and then stopped: "Installation failed due to error: -207" I subsequently attempted to unstuff the file with Stuffit Expander 6.5.1. It worked without any error messages, and the unstuffed installer worked flawlessly. Hope that's helpful.
JJ: did we upgrade the version of stuffit used to pack the archives since 6.x? Gregg: could this be the cause of all the -207 errors reported for the 7.0 beta? If so maybe we should go back to an older version of stuffit (what are the advantages of the newer version?) assuming the newer versions read the old format fine as is usually the case.
Assignee: dveditz → jj
Component: Installer: XPI Packages → Build Config
Keywords: nsbeta1
I think it is very possible that this is what we saw from the feedback. There were definitely a handful of -207 error reports. JJ - should we consider going back to the older version of stuffit?
The version of Stuffit Engine used on the build machines is 5.5 to ensure compatibility with all 5.x and 6.x versions of the expander. I don't think that's the problem. It's possible that archive corruption could occasionally occur during download or even during the build process, and Chris Johnson's comment #4 only tells me that a more recent version of Stuffit Expander handles and corrects the error better than an older version (6.5.1 vs. 6.0.1 in his case) Also, if I understand Chris' comment correctly, the corruption is in browser.xpi which gets created with another tool, Zipit, with which we've had intermittent timeout issues... I would lean towards this as the source of the problem. I can look into upgrading that tool to a more recent version as we have not been able to isolate nor reproduce the problem consistantly.
Status: NEW → ASSIGNED
I think it might be this known problem with Stuffit 6.0.1 (and only 6.0.1): http://www.aladdinsys.com/support/techsupport/mac/dlx/dlx6512.html I just tried unstuffing mozilla-mac-trunk-full.bin for build #2002062508 with Stuffit Expander versions 5.5, 6.0.1, and 6.5.1: 5.5 - expands and installs without any problems 6.0.1 - stuffit error message about browser.xpi and -207 error on installation 6.5.1 - expands and installs without any problems I'm sure my hard drive is pretty fragmented, so everything seems consistant with this technote. The connection with fragmentation also explains why some people (like me) consistantly get this problem while others never do. JJ: The problem is always only in browser.xpi; however, if I unstuff the full installer with 6.0.1, download an uncompressed browser.xpi from the mac-xpi directory of build directory, and then replace the damaged browser.xpi in the unstuffed installer's folder with it, I get a perfectly functional installer. I don't understand all the details here, but I think that suggests the problem isn't with the creation of browser.xpi, right?
> I think it might be this known problem with Stuffit 6.0.1 (and only 6.0.1): > http://www.aladdinsys.com/support/techsupport/mac/dlx/dlx6512.html Yep, sounds like exactly that. Mozilla's full installer is ~15MB, with 10Mb for the browser.xpi file alone. > that suggests the problem isn't with the creation of browser.xpi, right? right. The only thing to do I guess is to mention this point on the release notes or on Mozilla's web site download pages. Marking fixed, since the workaround is to use Stuffit Expander 5.x or 6.5.x, but NOT 6.0.1
Status: ASSIGNED → RESOLVED
Closed: 23 years ago
Resolution: --- → FIXED
Fixed isn't right, we didn't do anything... Adding relnote keyword. Should we close it INVALID? or wait until the relnote has been written.
Status: RESOLVED → REOPENED
Keywords: relnote
Resolution: FIXED → ---
To add a release note, please add a description to Bug 133795.
> To add a release note, please add a description to Bug 133795. done. Dan, we traditionally use FIXED when we have a workaround for a bug, which is the case here (use a different version of Stuffit Expander). If you prefer WORKSFORME or INVALID this time, I don't mind, but please close it back.
WFM with non-buggy version of Stuffit.
Status: REOPENED → RESOLVED
Closed: 23 years ago23 years ago
Resolution: --- → WORKSFORME
Verified WFM. Workaround is to use Stuffit Expander 5.x or 6.5.x, but NOT 6.0.1.
Status: RESOLVED → VERIFIED
Product: Browser → Seamonkey
You need to log in before you can comment on or make changes to this bug.