Three Monkeys, Three Typewriters, Two Days

It seems that every line of that code was meticulously documented. Of course, none of those comments may actually be "useful". (IANAP)(1)

(1) IANAP: I am not a programmer

Well, it is much better in v 1.33 that is true :-)

Peter: look at that version to see the difference between a plain comment that brings no clue and a good comment, you'll see it even if YANAP :-)

BTW, the whole comment for this method seems one character to the left in LXR...

Wow. Is that David Hyatt? I mean, THE David Hyatt, the same that is currently "leading Mac users into idolatry"?

OTOH, checking the log I see that he made the code a lot better very quickly. In fact he posted v 1.2 explaining how to call getCellAt 17 minutes after 1.2. Sure you can't compare that to leaving code in that state for years...

[PS: I am a Mac user, in case you wonder. Leave Dave alone!! :-) ]

Yes, that is THE David Hyatt. No, he didn't add any documentation for getCellAt. Documentation for getCellAt was added in April 2005 (3+ years after the interface was created), and not by David Hyatt.

And no, I have no plans to "leave Dave alone." Shoddy work is shoddy work, no matter how much you idolize the perpetrator.

The box objects were never considered to be real interfaces. They were just scriptable hooks so that the XBL glue code could communicate with the render object from JS.

While I won't defend the lack of documentation (all headers should be well-documented ideally), I will point out that the climate back then was such that I was working insane hours trying to do many many things at once.

Maybe that doesn't excuse it, I don't know, but I thought maybe you would appreciate my perspective.

David, I'm aware of the climate at the time. I believe that climate was a major management error on the part of Netscape, and that the Mozilla project has trying to recover from the effects climate for a while now, with only limited success, as we can see. A little bit less rushing back then could have meant a much better and more maintainable codebase for years now. :(

The "real interfaces" distinction doesn't really hold water, though. XBL glue code is exactly what's ended up buggy because the authors of said code didn't really know what the box object code did under the hood.

A little less rushing then would have meant there'd be nothing worth talking about now at all. We'd certainly have nowhere near the same level of infrastructure, for instance. It's quite possible we'd never have gotten the products far enough off the ground to get the level of funding or corporate involvement the project has received.

Who needs funding! Hippies onwards!

Ian, the corporate funding issue is one that certainly bears thinking about; I wasn't involved in that, so I don't know what the scene looked like. From a pure engineering perspective, I think a lot more time has been lost to this approach than gained by it.

There's a lot more to bear in mind in product development than just the engineering perspective. Especially before you publicly criticise someone's code in a way they might find offensive.

Quite frankly, the post was meant to be offensive. If nothing else, there was plenty of time after the crunch was over when things could have been documented. They were not. Instead, people kept hacking around the same old issues, adding more and more crud.

You were complaining about the initial checkin. That's what I was responding to.

(In anycase, the person who did the original checkin moved to different projects before the era of "plenty of time" arrived. I don't know who you had in mind for doing all the documentation.)