03:58
<MikeSmith>
zewt: wonder why they bother to use # for the PresentationFramework/src/Framework.. part
03:58
<MikeSmith>
zewt: that part doesn't look like a familiar fragment anchor at least
03:59
<MikeSmith>
zewt: and the second # has that problem of causing the URL to be invalid per the URL spec
04:01
<MikeSmith>
(also nice nostalgic use of frames there)
04:20
<zewt>
MikeSmith: they do it because that's client-side, not server-side
04:20
<zewt>
and the url spec is wrong, I think
04:20
<SamB>
bother that
04:21
<zewt>
(i don't even know what that means)
04:23
SamB
is just hopeful about higher-anchors happening at some point ...
04:59
<zewt>
hotmail represent
23:41
<zewt>
things that need to die: uuids as random tokens (just use a 128-bit random hex string, avoid a whole silly mess of formatting and bit twiddling)
23:43
<SamB>
(... won't that make our COM code slower to match those IDs? ;-P)
23:45
<zewt>
qq
23:46
<zewt>
just seeing them in URLs (stereotypically, microsoft URLs) and cringing
23:46
<zewt>
(they're also completely pointless to use as blob URLs and are just extra stuff to bloat specs, but since everyone's already using them for some reason... oh well)
23:46
SamB
is not sure if he's joking or not, but he's heard of optimizations related to consecutive GUIDs ...)
23:47
<zewt>
consecutive IDs aren't random tokens
23:47
<SamB>
I know
23:47
<SamB>
at least, clearly not 100% random
23:48
<zewt>
(though if you use a, say, 116-bit random token and use the rest as a counter, you probably still get enough randomness)
23:48
<zewt>
(v4 uuids are only 112 bits random to begin with)