| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10 11
<ijchain> <stevel> n
<ijchain> <stevel> woops
<ijchain> <rmax> Steve, need to focus yourself?
<ijchain> <stevel> indeed
<cleverly> "I had a question about the amount of data you are storing" (jcw)
<ijchain> <LucMove> sure
<ijchain> <GPS> yes
<ijchain> <torsten> yeah
<ijchain> <dkf> gwlester speaking
<ijchain> <torsten> oh!
<ijchain> <LucMove> *that* is his British accent?
<ijchain> <stevel> Louisiana
<ijchain> <dkf> As British as Hurricaine Katrina... :D
<ijchain> <LucMove> he pulled my leg saying he was British
<ijchain> <rmax> stevel: yesterday he promised to speak with his best british accent today.
<ijchain> <stevel> ah
<ijchain> <LucMove> right. that must be his best, then
<ijchain> <rmax> :)
<ijchain> <schelte> Writing the oo code already?
<ijchain> <stevel> looks like minesweeper ;)
<ijchain> <schelte> So that's what an object oriented system looks like!
<ijchain> <markroseman> objects? dammit, better not happen in tcl!
<ijchain> <kbk> cd ~ - probably back from there.
<ijchain> <stevel> what colour will the objects be?
<ijchain> <rmax> ... talking of coffee ...
<incith> starbucks is horrible :| second cup > starbucks. /me runs away?
<ijchain> <stevel> it's better than the hotel coffee
<ijchain> <dkf> There's no italian coffee house within a block...
<incith> Ah. Fair enough.
<ijchain> <markroseman> and my espresso machine is far far away.
<ijchain> <dkf> And anyway, starbucks do a good mocha
<ijchain> <dkf> their filter coffee's OK, but their mochas are actually good
<ijchain> <dkf> (OK == got lots and lots and lots of methylxanthines)
<ijchain> <miguel> objects should be mauve - they have more ram
<ijchain> <dkf> by that measure, objects should be coated in fleeces too ;)
<ijchain> <dkf> ram == sheep == covered in a fleece
<ijchain> <dkf> == awful pun :)
<ijchain> <stevel> sheared objects?
<ijchain> <dkf> :D
<cleverly> "do you run into a problem if you have over 1,000 clients, do you run into probles w/the # of open sockets" (jenglish)
<cleverly> "even 300-500, you haven't run into any os limits?" (jenglish)
<cleverly> "your system maintas long standing socket ocnnections between the lcients and servers, is that hours days weeks?" (don porter)
<patthoyts> udp would help with such limits.
<cleverly> "related to that do you have a problem if the client dies w/o closing the connections because tcp/ip connections will live for a long time if you dont have keepalive set?" (cant tell who)
<cleverly> "not if the machine dies and he hasnt sent a tcp/ip packet" (same)
<ijchain> <markroseman> must be about time for steve to start writing his slides
<xet7> http://www.tkgate.org
<xet7> there's screeenshots of app he's talking
<xet7> http://www.tkgate.org/fig/2.0/coke2.gif
<azbridge> *** Cameron entered
<ijchain> <Cameron> PT, I have two jobs that might interest you. Where/how can we chat?
<ijchain> <gwlester> Cameron, was that in the correct window?
<cleverly> on the price of power in a data center?
<cleverly> "did you have to--are you making an extension to the verilog language to tie into the vpd's?" (bob techentin)
<cleverly> "you need to have both tkgate and the verilog simulator then" (bob)
<cleverly> "do you have plans for synthesis?" (jorge lima)
<cleverly> "do you include the timing issues in your verilog simulations?" (d bazin)
<cleverly> "how did you create the graphics on the website?" (drh)
<ijchain> <stever> yay!
<ijchain> <LucMove> where is stevel from?
<ijchain> <stever> AU
<ijchain> <rmax> Australia
<ijchain> <GPS> LucMove: Perth Australia
<ijchain> <LucMove> string repeat thanks 3
<cleverly> "which of the various sha2 algorithms does it use when you specify sha2?" (drh)
<ijchain> <patthoyts> 256
<cleverly> "does cryptlib just have a default?" (drh)
<cleverly> "is the paper on the cd?" (someone next to don porter)
<ijchain> <rmax> http://tclers.tk/tcl2005/stevel1.jpg
<cleverly> "supports all the standard suspects?" (gwlester)
<cleverly> "did you take a look at [uninteligible?]" (oscar)
<cleverly> "is there any support for supporting ssl protocols or tls?" (oscar)
<cleverly> "how big is cryptlib?" (larry mcvoy)
<cleverly> "binary size" (larry mcvoy)
<cleverly> "i was just curious what kind of applications you would find using a cryptkit--i'm not doing banking, i suddenly envision this license manager that sends out encrypted emails about license [usage], what kind of capabailities do you forsee?" (bob t.)
<cleverly> "anything you'd distributed outside the firewall" (bob t.)
<cleverly> "do you know if it has rsa?" (joe m.)
<cleverly> "it has everything you'd ever want in crypto" (someone in the back)
<ijchain> <patthoyts> tcllib's crypto packages will ultimately use this as an accelerator btw.
<ijchain> <patthoyts> It already does for sha1 and md5 I think.
<cleverly> "how big is cryptkit?" (dkf)
<cleverly> "no I mean the compressed kit" (dkf)
<ijchain> <rmax> the chat is visible on the projector right now. So this is the chance for you out there to say hello to the local audience.
<ijchain> <torsten> hi folks! glad to be with you
<cleverly> "are we going to get any feedabck like you do on a microphone?" (bob t.)
<ijchain> <stever> heh
<cleverly> "do these things generally compress [first]?" (john h.)
<cleverly> "I know its easy, but is it done in cryptkit yet?" (john h.)
<ijchain> <patthoyts> openssl's ssleay32 is 640K static library (win32)
<ijchain> <patthoyts> the libeay32 part is 3M
<ijchain> <kbk_> Q: Would exporting a Stublib for libtommath in 8.5 help you?
<xet7> Can cryptkit help to secure the crypto keys and tcl code, securing itself?
<ijchain> <stever> q: can we easily pare down the cryptlib to create a custom version for small embeded apps that just contains a subset of functions?
<ijchain> <patthoyts> Use tcllib for small stuff.
<ijchain> <patthoyts> We have C impls for rc4 and one or two others.
<ijchain> <torsten> clap clap
<ijchain> <dkf> break for 11-12 minutes
<ijchain> <patthoyts> The tcllib critcl accelerator has C impls for md5,md4,sha1 and rc4
<ijchain> <rmax> stream down for the break
<ijchain> <aku> pat - One thing I noticed is that we have critcl code for base64, but it doesn't seem to be operational
<ijchain> <patthoyts> No - we dont yet.
<ijchain> <aku> OK. I wondered when I did comparative tests ... I thought that the C code was running and the times were the same ...
<ijchain> <patthoyts> The base64c file is the module file that we need to get the uuencode and yencode impls built
<ijchain> <aku> I really like the benchmarks ... You might have seen some work on tree, in the tcl impl., not the C side ...
<ijchain> <patthoyts> So far it doesn't contain a base64 implementation. Should do.
<ijchain> <aku> Interesting ...
<ijchain> <patthoyts> Yeah - the bench code is a much better solution.
<ijchain> <patthoyts> We can remove the timing files now.
<ijchain> <aku> Ok. I will do that over the weekend
<ijchain> <schelte> Thanks for relaying all that interesting stuff again today.
<ijchain> <aku> I can send you two tcl scripts which use this for comparison Tcl/C, and different Tcl sandboxes.
<ijchain> <aku> which _I_ use
<ijchain> <patthoyts> although one feature I like in tclbench that I will add back to these files is that you can source a tclbench file and it does a single run with the current interp. eg: tclsh tclbench/tcl/sha1.tcl does something like the timing scripts I did.
<ijchain> <aku> How is that done ?
<ijchain> <patthoyts> Possibly your sak options permit this kind of operation too. I havn't looked into it yet
<ijchain> <patthoyts> At the bottom it switches on [info command bench]
<ijchain> <aku> Actually my default is to look into the PATH if nothing is explicitly specified
<ijchain> <aku> We can change this to look at the interp used by SAK
<ijchain> <patthoyts> The problem I find is I have multiple tclkits and tclsh's lurking on my PATH :)
<ijchain> <aku> Yeah. So using the SAK nterp is the better default
<ijchain> <patthoyts> esp on windows I have tclkit84g.exe, tclkit84.exe, tclkit85a3g.exe ... ad nauseam.
<ijchain> <patthoyts> ye
<ijchain> <patthoyts> s
<ijchain> <aku> Ok, I will fix this in a minute ... I believe I have to fix normalize as well
<ijchain> <aku> If a cell in the ref-column is empty then the whole row should be empty, because we cannot normalize
<ijchain> <aku> Currently I keep the time data in the other columns, making for ridiculous factors for the unwary
<ijchain> <patthoyts> heh
<ijchain> <patthoyts> akf for a while.
<ijchain> <aku> ok, see you
<ijchain> <miguel> cd ~
<ijchain> <aku> Hm, we should be in the app show & tell now ?!
<xet7> no stream
<cleverly> still on break
<ijchain> <rmax> stream up again
<cleverly> though stevel is setting up
<ijchain> <aku> I have sound now
<ijchain> <aku> Thanks
<ijchain> <rmax> the break was 10 min later than planned.
Return to tcl
or
Go to some related
logs:
rock
osdev
rock
gaim
djgpp installer
|
|
