Fedora Core 3 and the Wireless Card that Refused to Cooperate

This is a bit of a retrospective post, I am revisiting an earlier part of my Linux journey.

At some point in late 2004 or early 2005, I fell out with Fedora Core 2.

Not permanently. More in the:

“I have learned enough from this particular flavour of suffering”

sort of way.

So naturally, I installed Fedora Core 3.

At the time I was documenting everything on a little website called LinuxJourney.com, which began as a place to store notes so that when I inevitably broke Linux and rage-quit back to Windows, future me would not need to relearn absolutely everything from scratch again.

This was before:

• Stack Overflow
• YouTube tutorials for everything
• cheap VPS environments
• disposable cloud infrastructure
• “learning in public” becoming a thing people intentionally did

Mostly it was just:

• forums
• HOWTO pages
• Google
• and increasingly desperate experimentation

The wireless networking problem

The thing I remember most about Fedora Core 3 was wireless networking.

Or more specifically:

my BENQ AWL100 PCMCIA wireless card refusing to cooperate with reality.

Modern Linux users may never fully appreciate how chaotic wireless networking used to be.

Today:

• the installer detects your hardware
• firmware loads automatically
• NetworkManager appears
• WiFi mostly works

Back then, hardware support felt more like a negotiation. Sometimes Linux recognised your device. Sometimes it recognised part of your device. Sometimes it recognised the device but not the chipset. Sometimes it recognised everything correctly and still refused to connect because the moon was in the wrong phase.

The confidence phase

I began the installation process optimistic. This was already a mistake.

Fedora Core 2 had eventually worked with this card, although not especially elegantly, so my assumption was that Fedora Core 3 would improve things.

Instead I discovered that:

• the wireless card list looked identical
• my card still was not properly supported
• and support for my hardware appeared to have advanced by approximately zero percent.

I wrote at the time:

“Some things might have changed in FC3, but support for my wireless card is not one of them.”

Still accurate.

Linux in the “compile everything yourself” era

At this point the internet began recommending:

• driverloader
• kernel modules
• linux-wlan-ng
• pcmcia-cs
• compiling drivers manually
• downloading source tarballs from random websites
• and, naturally, building kernel source

Which was interesting because I did not yet properly understand what kernel source was.

This led to one of my favourite rediscovered moments from those old posts, I could not find the kernel source anywhere on the system and genuinely believed something had gone catastrophically wrong because:

“I was under the impression all Linux distributions came with the source”

Which, in fairness, felt like a reasonable assumption at the time.

Eventually I discovered that Fedora no longer included it by default, meaning I now needed to:

• download the correct source package
• build it
• point modules at it
• compile against the correct kernel version
• and try not to destroy the entire system in the process

I remember thinking this all felt slightly ridiculous. I still think this.

Dependency hell was real

One thing these old posts reminded me of is how fragmented Linux desktop computing still was at the time.Package management existed, but it was rough around the edges.

Fedora shipped with:

• up2date
• yum
• RPM packages
• dependency chains from another dimension

At one point I became so annoyed with package management that I wrote an entire article about installing Synaptic, primarily because I was tired of manually chasing dependencies around the operating system like some kind of technical bounty hunter.

My frustration at the time was wonderfully specific. I described situations where:

• package X required Y
• Y required Z
• Z could not be found
• and therefore absolutely nothing useful happened

Modern package ecosystems are not perfect, but they are substantially less emotionally exhausting than they used to be.

The strange thing is that I loved it

Reading these posts back now, I expected mostly embarrassment. There is certainly some of that.

• The writing is over-explained.
• The tutorials are chaotic.
• Every problem becomes a saga.

But underneath all of that is something else:

• curiosity
• persistence
• experimentation
• and a growing obsession with understanding systems properly

I started wanting wireless networking. I somehow ended up:

• compiling modules
• learning package managers
• rebuilding kernels
• editing boot loaders
• troubleshooting ACPI
• mounting NTFS partitions
• installing Apache
• running forums
• managing Linux hosting
• and accidentally teaching myself infrastructure concepts years before I would have described them that way

That part feels surprisingly familiar.

LinuxJourney.com

The other thing I had forgotten was how quickly LinuxJourney stopped being “just notes”. At first it was static HTML uploaded manually to shared hosting. Then:

• forums appeared
• articles multiplied
• users registered
• people started helping each other
• and I slowly moved from documenting problems to supporting other people through theirs

Which is honestly a slightly strange thing to realise twenty years later. Especially because I was still learning almost everything myself. The hosting itself now feels historically interesting.

• This was not a VPS.
• Not cloud hosting.
• Not containers.

It was one of those early Linux hosting environments where you got:

• FTP access
• PHP
• MySQL
• limited storage
• and just enough control to break things accidentally

At the time it felt incredibly advanced.

Looking back now

What strikes me most revisiting all of this is how much of my current thinking already existed in rough form. I was already:

• documenting as I learned
• trying to reduce friction
• building systems around recurring problems
• teaching while still figuring things out myself
• and becoming increasingly interested in infrastructure rather than just applications

The technology has changed massively since then. The habit of turning relatively small technical annoyances into disproportionately large side projects appears to have survived completely intact.