I compile my own AMP stack from source, so my setup is a little different than most.

On my laptop, apache and php both live in /apache2, so my php binary is at /apache2/php/bin/php, and php.ini lives in /apache2/php/lib/php.ini

Here’s what I did to make xdebug work:

# cd /apache2/php/bin

# pecl install xdebug

Which gave me a shiny new xdebug.so in /apache2/php/lib/php/extensions/no-debug-non-zts-20060613

Pecl then lied to me, and said I should add “extension=xdebug.so” to my php.ini

That didn’t work.  I got message like:

PHP Warning:  PHP Startup: Unable to load dynamic library ‘./xdebug.so’ – (null) in Unknown on line 0

in apache’s error_log

What did work was adding the following line to php.ini, and restarting apache:

zend_extension=/apache2/php/lib/php/extensions/no-debug-non-zts-20060613/xdebug.so

Now, this directly contradicts what my default php.ini said about not using directory information in the extension line, but fiddling with extension_dir didn’t seem to work for some reason.

Anyway, this worked for me.

Worth noting.  I still get some junk in apache’s error_log that looks like:

“Failed loading xdebug.so:  (null)”

but xdebug appears to be functioning properly (I get pretty-printed var_dumps, and the profiler will write cachegrind files just fine)

It simply picks a random quote from a unix fortune cookie file, and inserts it into your wordpress template.

I wrote it when I created this blog, to enable that random tagline you see in the top-left, under the blog title.

For More Information, visit the GoodFortune Home Page

Or just download the GoodFortune Plugin now.

The hideous rainstorm over eastern PA, NJ, and NYC defeated our dream of making the whole trip in 24 hours door-to-door.  Ended up taking more like 27.  The storm was pretty intense, but the car (and the fancy tires it came with, I guess) did really well.  Lucky it came with good tires and wipers, otherwise we’d probably have had to stop and wait out the storm — that’s how bad it was.

Now to go check out the digs Zsuzsa has hooked up for the short-term in the city, have some supper, and generally reunite….

So I guesst his means I’m once again a New-Yorker.

It was a bit funny, watching the schism between product-centric people like Jason and client-services people (like most agency people) flare up.  However, I find that a lot of what Jason talked about was what I tend to agitate for when starting new client relationships.  Client-services people are generally terrified of saying things like “We think we can do something great for $X, let’s get started”.  My thinking is that if you do good work, and are worth what you’re getting paid, this can serve as a great filter for clients and projects.  Clients that want everything specced out, discussed, wire-framed, etc, before they open their checkbook are just asking for free work.

I managed to get a decent audio recording on my laptop, and took some notes.  Here are the notes and audio.  Perhaps I’ll add more commentary later:

Audio: jasonfried_20080731.mp3

Notes:

NDAC Presentation – J. Freid.
7/31/2008 @ Oriental Theatre in in Denver

=============

Planning is overrated: Start doing stuff

Don’t let planning scare you out of doing something

Blah blah blah, sharing/teaching/open-sourcing is good… blah.

Interruptions bad.  Open-floorplan offices contribute.  When it’s easy to interupt somesome, you break their time up into smaller and smaller chunks.

Passive collaboration techniques can help with interruptions.

Big companies wish they were small.

Err on the side of simple.  37s’ big mistakes were trying to do too much.

Focus on things that don’t change.  Let your competitors thrash about and “compete” with each other to do “something new”.

Fancy PM charts/documents etc create the “Illusion of agreement”

QUESTION: How do you figure out ROI possiblity?  How to determine if a client (for those of us who do client work) can afford to finish what they start?

Jason claims that they won more work when they got rid of abstractions and started building stuff (or something).

Make dirty comps.  Things don’t need to line up.  Don’t waste time on things early on.  Designers want to be perfectionists, but they’re stupid assholes.

Hire good writers (duh)

Meetings are toxic.

Need, Can’t, and Easy are BAD WORDS.  When you say easy, you’re always talkinga bout someone else’s work.  (QUESTION: I say things are easy all the time, because I see a simple solution — but easy combines with “need” for some idiotic feature.)  You think you need something to launch, or can’t launch without something.  This is just limiting, etc.

Decompose problems into the smallest atomic things possible.  If you can solve 1/4th of a problem in a few mintues, do that, instead of spending days solving a whole problem.

========
Q&A
========

How to apply all this shit to client-work?

- Promising stupid deliverables all along the way is dumb.  Clients don’t are about that.  They want to see stuff done.  How much will it cost, when I can I see it, etc?

… the best frameworks are the ones extracted from real projects.

- When customer/client proposes a solution, they’re really describing a problem, your value is that you might have a better/simpler solution to their problem.

Favorite website?  The drudge report – it’s clear, one page, etc.  You can scan it all in 20 seconds and get it all.  Compare it to CNN, MSNBC, NYTimes, etc.

Despite free stuff, people who usually pay are paying from day1.

Backpack started out as free-first, upgrade later.  Offering the option to pay right from the start improve revenue huge.

Roles @ 37 signals.:
1 cust service
1 jason (biz and design)
1 full-time designer
1 full-time writer
5 people are programmers

5/5 chicago/elsewhere.

Duffy design in minneapilis — their site just shows pictures of their work, that’s it.  Jasont thinks that’s cool.

Internal disagreements:  (ultimately, J will decide, but that never happens).  Generally, whomever is more passionate about something, they get to hold the bag/take responsibility.

After I suggested to some folks on the nyphp “talk” list that compiling your own daemons from scratch isn’t painful so long as you maintain some scripts to do it for you, Dan Horning called my bluff.

So instead of just dumping some stuff on a server, I went all out and created a blog (since I have some other code I want to publish sooner or later, anyway).

This stuff is not a “product” of any sort.  It’s just an example for anyone interested in building their own apache/php stack from source, with openSSL, and various other goodies.  These scripts cand be a good starting place for building your own.

Point of Interest: These build scripts built the Apache and PHP binaries that are powering this blog.

NOTE: I have no idea what kind of damage this might do to your system.  Please don’t run this stuff straight out of the box on any machine that is not ready to be rebuilt from scratch.  Really, I mean it.  The best thing you could do is look at what I’ve done and build your own scripts to suit your situation.

Anyway, since it’s getting late, and I want to get this stuff published before hitting the hay, you can download the scripts by themselves here:

• http://blog.electricmindcontrol.net/files/build-php5-apache22-openssl-0.1.tar.gz – 31M – the scripts, plus the source tarballs they need.
• http://blog.electricmindcontrol.net/files/build-php5-apache22-openssl-0.1-slim.tar.gz – 10K – just the scripts.  You’ll need to go hunt down the various tarballs yourself.  It’s fun!

If there’s interest, it might be fun to actually try to productize this a bit.  My approach is pretty distro-agnostic.  For instance, I compile all this stuff into /php5web, outside of the host system’s setup.  I don’t really see other people doing this, but it’s kind of nice.  I typically run servers with 2 or more distinct stacks (various permutations of apache 1.3.x/2.0.x/2.2.x and php4.x/5.x) listening on different IPs — which is pretty handy sometimes.

One last hint: apache is kind of dumb about openssl.  You might need to add /<yourinstalldir>/openssl/lib to your /etc/ld.so.conf and then run /sbin/ldconfig if apache complains about not finding libssl.0.9.8.so or somesuch.

Finally.

After having created various blogs for various people, I finally broke down and made my own.

I probably owe some people some money now.  And that hat is looking mighty tasty.

If we want to blame someone, it’s probably Dan Horning — an otherwise great guy.  I got uppity on a mailing list and offered to share something from work (some handy scripts I wrote to compile the AMP part of LAMP on linux (CentOS 5, in particular).  He thought sharing my stuff would be a great idea.  What villany, playing to my pride like that.

Anyway, with any luck, I’ll actually get some code posted here in the next few hours.

In the meantime, welcome to “Electric Blog Control”