43394 - addDirectory: Certain case returns -226 (FILENAME_ALREADY_USED)

Status: VERIFIED WORKSFORME

(Core Graveyard :: Installer: XPInstall Engine, defect, P3)

Description

[Jimmy Lee]

Build: 2000-06-20-08-M17(MAC)

1. From http://jimbob/trigger3.html, select f_adddir_regname_absolute from 
   Functional drop-down menu
2. Click Trigger case button
3. Click Install button from confirmation dialog

RESULT:
Directory is not installed.  Install.log returns error -226.  Windows and Linux 
behave as expected with no errors.

EXPECTED RESULT:
Directory is installed with no error -226.

INSTALL SCRIPT:
///////////////////////////////////////////////////////////////////////////////
// f_adddir_regname_absolute.js
// 
// Functional test
// 	• Creates subdirectory "f_adddir_regname_absolute" under "Program" and 
//        installs "thefolder" directory which contains smrtupdt.txt
//
// ~ XPInstall QA
//   August 13, 1999
//
//  March 7, 2000  jimmylee  Renaming test case to less than 31 chars to                       
//                           support Mac OS
///////////////////////////////////////////////////////////////////////////////

var regName = "f_adddir_regname_absolute";
vi = "1.0.1.1";
var jarSrc = "thefolder";

initInstall("Functional: f_adddir_regname_absolute", "adddir_regname_absolute", 
vi, 0);
f = getFolder("Program", "f_adddir_regname_absolute");
err = addDirectory("/" + regName, vi, jarSrc, f, jarSrc, true);
logComment("addDirectory returns = " + err);

if (0 == getLastError())
	performInstall();
else
	cancelInstall();

Comment 1

[Samir Gehani]

-226 = FILENAME_ALREADY_USED

It appears that the only time this error is thrown is when a file already exists 
with the same name as the folder you are trying to create.  Is this the case on 
your filesystem?  Also, are you experiencing this on "install" or "update" -- 
i.e., when you install the file for the first time or when you install it by 
running the .xpi for the second time (or the file already exists)?  

Comment 2

[Jimmy Lee]

The file/directory does not exist prior to triggering the test.  This error 
occurs when "installing" or triggering for the first time.

Comment 3

[Samir Gehani]

Can't reproduce on my debug build or by clicking on the xpi (since triggering is 
broken) using today's (June 21) optimized Mac commercial build.  

However, I'm seeing this on Jimmy's machine on his freshly installed (June 21) 
optimized Mac mozilla build.  Furthermore, when I use the identical build on my 
machine (i.e., June 21 optimized Mac mozilla build) I don't see theproble: 
installation and replacing both take place just fine.

While attempting to isolate the problem I rebooted Jimmy's machine only to find 
the Finder unresponsive -- took "some time" after reboot to respond to 
mouseclicks.  This leads me to believe this is a machine specific issue.  In 
fact, I tested this (same mozilla build) on Sean's Mac OS 9 G4 (identical to 
your configuration) and the problem did not manifest itself.  

Jimmy,
Can you try and reproduce this on a another machine?  Also, can you try 
adjusting the memory given to the mozilla app?  And of course, taking up the 
always unpleasant "reinstall your OS" please recommendation would certainly help 
:o)  

I am strongly leaning towards marking this worksforme.  If this is in fact 
reproducible (like if this a thread safety issue that is inconsistently 
reproducible) then it would be a terrible bug that could manifest for users of 
the installer in which case we should try and track down a fix for nsbeta2.  
Until we get stronger reproducibility I doubt if it'll be blessed as nsbeta2+.

Comment 4

[Samir Gehani]

Oh, one curious anecdote: the Mozilla Installer itself that used the very same 
xpinstall.shlb worked fine (since that's how the build was installed in the 
first place).  The .xpis use the identical addDirectory() API.  Makes me less 
and less worried about this bug.

Comment 5

[Jimmy Lee]

I cannot reproduce the problem on another machine.  I am still reinstalling my 
OS and debugging.  I can still reproduce the problem on my machine.  
Investigating...

Comment 6

[Samir Gehani]

If after doing a low-level format of your hard drive this problem persists, we 
can spend a few minutes using a debug build with symbols to figure out exactly 
why this going wrong and confirm that it is in fact a corrupt 
filesystem/firmware issue as currently suspected.

Comment 7

[Jimmy Lee]

Build: 2000-06-26-08-M17(MAC)

After reformatting my entire drive and reinstalling the OS, I can no longer 
reproduce this error.

Comment 8

[David Epstein]

this can't be reproduced.

Comment 9

[Samir Gehani]

Per QA this is non-issue.

Comment 10

[Jimmy Lee]

Marking Verified.