Of FireFox history.dat ... and Mork ... and pathological culture
posted Friday, March 09, 2007 at 3:05 pm MSTBug 223476 – large history makes Firefox painfully slow in several ways
The way things go leaves me disheartened ... gagging.
I just revisited Bug 223476 – "large history makes Firefox painfully slow in several ways" in MozBugzilla, with ref to history.dat ... working through a text version (The "Dork" conversion utility is now hard to find; a copy of it is on Bug 242438.) I found that barely 1/5th of stored data could be regarded as "history information" ... masses of gack, produced by an ill-conceived and poorly implemented system. Bahh, humbug!
I wonder, does anyone remember History in NS3? I used it constantly ... it was, for me, an adjunct to blogging. Trim by deleting, sort variously ... the file was a wealth of information, not just data.
history.dat? Gack ... data, data everywhere, and not a thought to think.
I have managed in the past to delete some of the worst blocks of crap (huge URLs for sites not worth remembering), but it's always a crap shoot.
To quote Jamie Zawinsky (which I do while holding my nose), "it uses Mork, which is -- and I do not use these words lightly -- the single most braindamaged file format that I have ever seen in my nineteen year career."
If someone had documented Mork somewhere, or made their documentation discoverable, I think we would have improved this long ago.
(Some folk get paid. When I got paid I didn't just file things in a bottom drawer.)
p.s. I just ran "Dork" and my 2.5M .dat ... the text file was 1.29M
Deleting just the links that were comprised of google.ad gack reduced that to .62M ... a quarter. Deleting other gack (like adclick) reduced it to .49 ... i.e less than 1.5th!
We have lost NSGold functionality and created a gack collector.
BTW: "Re: High CPU with Beagle Thunderbird" reads in part, "the mork file parsing is the most expensive operation probably, and the most likely to loop incorrectly." Seems to me that Mork sux. Maybe that's just because of how it's been implemented. Or maybe it was just a matter of "anything but JWZ".
I just wrote an email to someone (un-named here) who's deeply implicated in all of this:
I just found a document on Mork ... finally.
It your Primer Text" from March of 1999.
I find this an interesting story ... Mork is implimented, and tens / hundreds of thousands of people never learn to use their Histories as really useful tools, the way we could with NS3.
You know, a lot of people have talked about Mork ... nothing good has been said. A lot of people have filed bugreports ... nothing has been done.
It amazes me that nobody has simply explicated the thinking behind the code that actually exists. And it leaves me disheartened.
truly
benp.s. When I was 19 I was flowi into the Arctic to take over maintenance of the SAC/NORAD-DewLine system. Nobody knew how it worked ... I had to figure it out as I went. I got $7k/month + isolation pay. That was 1974. I don't work anymore ... can't handle being grossed out by the way things are done. Who loses? "Who cares!" *sigh*
p.s. I just now wrote and ran a TextPad macro to strip ICON data out of my bookmark file ... the 589K original is now less than 150K ... almost 75% GACK!
