Closed Bug 74522 Opened 24 years ago Closed 23 years ago

installer should use extracted xpi file instead of local xpi when installing

Categories

(SeaMonkey :: Installer, defect)

x86
Windows 2000
defect
Not set
normal

Tracking

(Not tracked)

VERIFIED FIXED
mozilla1.0

People

(Reporter: bugzilla, Assigned: dveditz)

References

Details

Attachments

(1 file)

I have the following scenario (while testing another bug):
I've extracted an mozilla-win32-installer.exe file in a empty c:\mozinst 
directory, by using the -u parameter.

I then currupted the browser.xpi file so that I get an error when starting 
setup.exe

Then I downloaded a newer mozilla-win32-installer.exe file from mozilla.org and 
put that file into the c:\mozinst directory. I didn't extract it.

Now when I start the mozilla-win32-installer.exe it still complains about an 
corrupted browser.xpi file.

When using the mozilla-win32-installer.exe shouldn't just use/test the 
extracted xpi files and not those located in the same directory as the mozilla-
win32-installer.exe file ?
this is as designed for CD purposes.  On a CD, we can have the self-extracting 
.exe (that does not contain any .xpi files) and the .xpi files along side it.
Status: NEW → RESOLVED
Closed: 24 years ago
Resolution: --- → INVALID
This also makes being able to re-use the saved downloaded archives work -- 
otherwise what would be the point of saving them?
All you points makes no sense to me, since I talking about having a 
mozilla-win32-installer.exe that contains all the .xpi files.

Ok once again:
since I use "mozilla-win32-installer.exe" that contains all the .xpi files and 
these xpi files are extracted to temp place there's no need testing xpi files in 
the same dir as the "mozilla-win32-installer.exe" file. That was just my point.
This doesn't have anything to do with be able to run "setup.exe" and having xpi 
files in the same dir.
Maybe this should be WONTFIX instead of INVALID then. The easy workaround is 
to make sure you put the installer in a different directory if you don't want 
to use the downloaded .xpi files. Sean is moving on to the mail team and only 
has time to fix serious problems that will affect lots of people.
So there will be no fix to the remaining installer bugs?
I didn't say that, I said he "only has time to fix serious problems that will 
affect lots of people." This bug isn't one of those.
Ok.. cool. The installer is also 99% finished, while the mail could do with some 
help...
verified invalid
Status: RESOLVED → VERIFIED
reopening.
Status: VERIFIED → REOPENED
Resolution: INVALID → ---
reassigning to default owner since ssu's no longer doing installer work.
Assignee: ssu → syd
Status: REOPENED → NEW
what's going on here? closing, reopening a lot of times..
Target Milestone: --- → M1
Target Milestone: M1 → Future
I came across this bug today - I downloaded the full installer into the same
directory where I had previously downloaded the netscape 6 stub installer - so
there were a load of xpi files in the directory as well. When I tried to install
mozilla then it installed parts of netscape 6 as well as parts of mozilla 0.9.9.
I could understand it doing this if I had used the mozilla stub installer, but
when I used the full installer it doesn't make sense (to me) to use the wrong files.
*** Bug 127044 has been marked as a duplicate of this bug. ***
nominating... this leads to all kinds of wierdness that's generally hard to
diagnose.
Keywords: nsbeta1
*** Bug 116536 has been marked as a duplicate of this bug. ***
Dan talked to me about it.  It could be a simple as a two line patch to
nsinstall.cpp
I'll take this one since I've already dug into the cause and a possible solution
Assignee: syd → dveditz
Status: NEW → ASSIGNED
Target Milestone: Future → ---
Could I get an r= from curt or ssu?
Using this patch sysadmins who need the override behavior can still get it by
using the "-a <alternatepath>" option, but people who don't explicitly choose to
do this won't get messed up by accidentally having old cruft laying around.
Attachment #75425 - Flags: review+
Comment on attachment 75425 [details] [diff] [review]
Don't automatically send -a argument to setup

r=ssu too
nominating mozilla1.0 since this appeared to bite several people in the 0.9.9
milestone release and it's a simple, safe change.
Keywords: mozilla1.0
Target Milestone: --- → mozilla1.0
Attachment #75425 - Flags: superreview+
Comment on attachment 75425 [details] [diff] [review]
Don't automatically send -a argument to setup

sr=mscott
Comment on attachment 75425 [details] [diff] [review]
Don't automatically send -a argument to setup

a=asa (on behalf of drivers) for checkin to the 1.0 trunk
Attachment #75425 - Flags: approval+
Fix checked in
Status: ASSIGNED → RESOLVED
Closed: 24 years ago23 years ago
Resolution: --- → FIXED
verified on Mozilla/MachV
Status: RESOLVED → VERIFIED
QA Contact: bugzilla → gbush
Product: Browser → Seamonkey
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: