I found that Zend Studio had some strange issues when starting up (it didn’t) and the answer was to get the original OSX 10.5 Java 1.5 and put it back where it could be used and do some finagling with settings to let applications use it. It’s all spelled out in this article on how to fix Snow Leopard’s Java Problem

After this past friday’s software update though, it happened again. After a little bit of research and some trial and error, I found out that I had to re-do a part of the previous fix.

Specifically, I had to re-update the symlink for 1.5.0 in /System/Library/Frameworks/JavaVM.framework/Versions so it was pointed at my 1.5.0-leopard folder. I’m not sure if I’ll need to update my other 1.5 related folder to not point to the CurrentJDK.

My /System/Library/Frameworks/JavaVM.framework/Versions directory now looks like this:

1.3        -> 1.3.1
1.3.1
1.4        -> CurrentJDK
1.4.2      -> CurrentJDK
1.5        -> CurrentJDK
1.5.0      -> 1.5.0-leopard
1.5.0-leopard
1.6        -> 1.6.0
1.6.0
A
Current    -> A
CurrentJDK -> 1.6

All of these were updated on Dec 11 when I did the update. So a quick

sudo ln -snf 1.5.0-leopard 1.5.0

And things were good.

Using the built-in bookmark manager was useless since there seemed to be an underlying problem with the SQLite database Firefox uses, so I tried opening up the database using the excellent SQLite Manager Add on for Firefox. By running the PRAGMA integrity_check command I saw that the database disk image is malformed. Uh oh.

It had to be rebuilt.

First, I had to make sure I had a SQLite Command Line Client because I wouldn’t really be able to work on Firefox’s databases from inside Firefox.

Then I navigated over to where my places.sqlite database file inside my Firefox profile (which, for me was ~/Library/Application Support/Firefox/Profiles/{profile_name} )

I ‘connected’ to my SQLite database

sqlite3 places.sqlite

I reran the integrity check to see if I could garner more information

PRAGMA integrity_check;

Which told me some more information that I didn’t know about

On page 11 at right child: 2nd reference to page 1609
rowid 912 missing from index moz_bookmarks_itemlastmodifiedindex
SQL error: database disk image is malformed

I’d read that doing a straightforward dump/reload of the data should fix this sort of problem, so exited out of the SQLite shell and did a dump from SQLite

sqlite3 places.sqlite .dump > places.sql

Followed by load into a new SQLite database

sqlite3 places.new.sqlite < places.sql

But then I got a bunch of primary key errors

SQL error near line 872: PRIMARY KEY must be unique
SQL error near line 877: PRIMARY KEY must be unique
SQL error near line 883: PRIMARY KEY must be unique
SQL error near line 884: PRIMARY KEY must be unique
SQL error near line 885: PRIMARY KEY must be unique
SQL error near line 888: PRIMARY KEY must be unique
SQL error near line 889: PRIMARY KEY must be unique
SQL error near line 890: PRIMARY KEY must be unique

I popped open places.sql and checked out the lines in question. I saw what appeared to be the primary key which had stopped logically increasing and was sort of random

INSERT INTO "moz_bookmarks" VALUES(915,1,16546,2,110,'Bookmark Page Name #1',NULL,NULL,1256413281706712,1256413281711799);
INSERT INTO "moz_bookmarks" VALUES(29,1,33509,2,135,'Bookmark Page Name #2',NULL,NULL,1258662433785621,NULL);
INSERT INTO "moz_bookmarks" VALUES(28,1,31347,27,0,'Bookmark Page Name #3',NULL,NULL,1258472420957472,NULL);
INSERT INTO "moz_bookmarks" VALUES(27,2,NULL,2,134,'Bookmark Page Name #4',NULL,'',1258472420951405,NULL);

After a cursory glance through the rest of the file, I figured that I could change those primary keys to whatever I wanted as long as they were unique. A quick run through with that, running the same code as before

sqlite3 places.new.sqlite < places.sql

And success! I renamed my new places.new.sqlite to replace my old places.sqlite, restarted Firefox and everything worked out hunky dory.

A few notes:

• Make sure Firefox is completely closed before attempting to do anything with the databases from the command line. You will get errors about the database being locked
• After restarting Firefox, I did notice issues with my AwesomeBar not quite remembering my usual auto-complete and I also had to click through all of the items in my toolbar in order to get the favicons showing again, but so far nothing major

time

This lets you do a quick and dirty profile on some code you’re executing.

Basic Usage

time <command>

This will execute the command and put out a summary detailing how long the command took in total, how long it was tied up in actual execution, and how long it was tied up in system overhead:

real	0m0.031s
user	0m0.004s
sys	0m0.023s

Another nifty function that does one thing and one thing only, but does it well. Thanks Unix philosophy!

Lifted from http://www.ampsoft.net/webdesign-l/WindowsMacFonts.html

I have an SVN repository cloned locally. I’ve made a couple of changes in order for my application to run locally and I committed those all against the “dev” branch I set up.

I then go off and do some actual work, which then needs to get sent back upstream to SVN. This is my workflow.

Starting in the dev branch, with all of my changes committed…

1. Because I should be making sure I have the newest stuff from SVN, just in case

   git svn rebase

2. Switch over to the master, which should mimic SVN

   git checkout master

3. Bring master up to date with the dev branch

   git rebase dev

4. Remove the commits I don’t want to push back to SVN

   git rebase -i git-svn

5. Because I’m paranoid and want to make sure only the commits I want are actually coming through

   git svn dcommit --dry-run

6. Commit it off to SVN

   git svn dcommit

Changing History

git rebase -i HEAD~37

Undoing crazy rebases

1. Find the commit you want to bring back, copy it’s sha1 starter

   git reflog

2. Boom, commit is back.

   git cherry-pick <sha1 starter></sha1>