[coreyh14444]

I was deep into OS explorations in this period. I ran BeOS, Yellow Dog linux on PowerPC, had all flavors of NT, etc.

My feeling about OS X at this time was that it was the right choice but that it was "too late" and that Apple took too long and the window of opportunity had passed.

I take that now as startup advice: It isn't too late!

[sanderjd]

I need to remember to remind myself of this more often!

The one I remember so clearly is back when I was using Pidgin (or Adium I think was another thing I used?) with a few different messaging apps, and my company was trying to figure out how to make chat work, maintaining an irc server that we of course couldn't get the non-nerds to use, looking at HipChat and a couple others that seemed really expensive for their feature set at the time. Basically none of this stuff supported mobile well.

This was 2008, maybe 2009. I remember thinking "this all sucks, but it's waaaay too late to make a chat app; I've been using IM apps for over a decade, totally saturated and commoditized".

But then I watched the rise of WhatsApp, Facebook Messenger, and Slack, all of which were tiny or non-existent at the time, and became gigantic businesses.

[code_duck]

I remember ICQ and Aim. It's not even clear to me why they died and were replaced, but that sure did happen.

[lesuorac]

AOL actively wanted AIM to fail.

https://mashable.com/archive/aim-history

[sonicanatidae]

ICQ is used a great deal outside the US. Aim, well that's AOL, so reason enough.

[scarface_74]

Isn’t that the ultimate survivorship bias?

[Macha]

In terms of who the winners were, sure. There were many also-rans like Viber or companies that won only in certain markets like Line or Telegram.

But the idea that the network effects couldn't be overcome was definitely false. Not only are MSN, AIM and YIM so dead their servers are shut down, but the services which displaced them (Facebook Messenger and Skype) have themselves been displaced by a third wave of services (WhatsApp, iMessage, Discord, Slack)

[cguess]

Viber is still big in certain places like the Philippines.

[pjmlp]

And several Mediterranean countries, it is easier to reach someone on Viber than WhatsApp.

[sanderjd]

My point wasn't about the specific companies that "survived" chat, but that any new chat products got huge after 2008 at all. From my perspective back then, I thought the whole category looked run out. But it had actually barely even gotten started, if you compare the number of users of chat apps back then to the number now.

[tambourine_man]

OS X was a bet on the future, that double buffering everything was OK because graphic cards would get there eventually. They certainly did, but we suffered through menus taking a good fraction of a second to drop down for a few years. I can still feel the pain.

I would never have taken that bet. I want computers to be fast now. It was the right bet.

[tambourine_man]

Another hard decision that was the right choice in retrospect was going with KHTML. I remember watching the keynote and waiting for them to announce it was based on Gecko. Camino had shown it was possible and worked well. Apple had hired David Hyatt. It was obvious they would go with Gecko.

There was an audible gasp when they announced it was based on KHTML. What the %^&$ is KHTML?

[tambourine_man]

Yet another one was going with LLVM.

Now that they had a decent running GCC on PowerPC they want to build a compiler infrastructure from scratch?

Apple is full of those decisions. I'm just glad they don't take advices from me.

[tinus_hn]

LLVM is because they don’t want GPLv3. They just went ahead with a stack where they can be in control on their own terms.

Basically GNU gambled and lost, they wanted to make a big stand against software patents and some other powerful forces had different opinions.

[GeekyBear]

> LLVM is because they don’t want GPLv3.

LLVM and Clang are also about having a compiler stack that provides the information you need for a modern graphical IDE.

For instance, GCC had made some decisions that made it difficult to point to which line of code caused a particular error message.

Apple's Chris Lattner covered this in a 2007 tech talk he gave at Google introducing LLVM and the (then) new Clang project.

https://youtu.be/VeRaLPupGks?t=1133

[ori_b]

> LLVM and Clang are also about having a compiler stack that provides the information you need for a modern graphical IDE.

Is that why all GNU utilities shipped with OSX were also stuck at the GPLv2 versions and never upgraded?

[kitsunesoba]

GPL v3 is part of it, but as I understand, back when the decision was made GCC was rather inflexible compared to LLVM/Clang which was also a big factor.

[Longlius]

I believe LLVM was chosen because it was easier to make new compilers to work with it. GCC's intermediate representation was a bit of a mess for a long time and from what I understand, Apple did not find GCC's Objective-C support to be satisfactory.

[mproud]

The venerable KHTML project was just officially discontinued this year.

[andrekandre]

  > double buffering everything was OK because graphic cards would get there eventually.

and they showed it in the very next version 10.2 jaguar

https://www.youtube.com/watch?v=4OaRF9W1bJk&t=3m33s

[tambourine_man]

I remember being very relieved that my card qualified for quartz extreme. Nice demo.

[freedomben]

I remember feeling that way too but had forgotten about it! Wow, what great life lesson I nearly missed, thank you! It's never too late (well, usually not)

[scarface_74]

Are you sure that was the right lesson to take for a startup?

Apple had millions of users, an already profitable business (by 2001 barely), and marketing reach that was far outside its market share.

They had the best marketer and product guy in the business for a CEE and one of the best logistics experts for a COO.

A startup probably has none of those things.