| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10 11
<ijchain> <kbk> Interesting concept - "developer's workshop" concurrently with tutorials. Packaging would be a particularly interesting thing to work on.
<ijchain> <kbk> In past years there has been talk about doing a track at Freenix.
<ijchain> <CoderX2> lots of things would be interesting to work on in that context
<ijchain> <CoderX2> TIPs, complex bug fixing, etc
<trebor_dki> if i use an regular expression in regexp $mask $string i get a result of 1, if i use the same regular expression in switch - the corresponding body is not executed - where are the normal pitfalls?
<ijchain> <CoderX2> i'm referring to the "implementers meetings" which it seems is different from what you are referring to, kbk ?
<ijchain> <kbk> Joe: The stuff that particularly interests me from this standpoint is the stuff that collocation really helps with - like getting stuff working x-platform - where people can *bring* their favourite boxes and be in the same time zone.
<ijchain> <CoderX2> yes
<ijchain> <CoderX2> i would like to see more group activities... "field trips", etc...
<ijchain> <silas> uhm... everytime a children of . is destroyed (I'm using [destroy [winfo children .]]) [bind . <Destroy>] is lauched. Possible to avoid this?
<azbridge> <suchenwi> trebor_dki: could you show a simple example?
<ijchain> <kbk> silas - The easiest thing is to compare %W in the <Destroy> binding with the window that you bound.
<trebor_dki> suchenwi: i have to do copy & paste via pen & paper, so it may take a while ... where do i have to post it to?
<ijchain> <kbk> The issue is that the default set of bindtags on any window $w is [list $w [winfo toplevel $w] [winfo cl*** $w] All]
<ijchain> <kbk> ... so the <Destroy> on a subwindow is paid to the toplevel.
<ijchain> <silas> uhm... all right... I'll do it with %W, thank you.
<azbridge> <suchenwi> I'd expect switch -regexp to honor the same conditions as regexp
<ijchain> <stever> the commitee can definitely schedule the time slots
<ijchain> <kbk> someone repeat what Gerald just said?
<ijchain> <tclguy> mumble mumble mumble
<ijchain> <tclguy> beer beer beer
<ijchain> <tclguy> mumble umble mumble
<azbridge> <suchenwi> trebor_dki: ideally, distill the problem into a one-liner and paste it here...
<ijchain> <silas> Kevin, it worked well, thank you.
<ijchain> <stever> next week
<ijchain> <stever> last week :)
<ijchain> <stever> I will be strick on the network resources
<trebor_dki> suchenwi: set m {.*\(strcmp\(ch, \"(.*)\"\)==0.*(DKI0001_.*)=.*}; set t "else if (strcmp(ch, "stahl")==0 DKI0001_const.stahl=d; if i do puts [regexp $m $t] i get 1; if i do a switch with $m as a pattern its body is not executed ...
<ijchain> <tclguy> beeeeeeeeeeeeer
<ijchain> <tclguy> sushi!
<ijchain> <kbk> 100 is exactly the wrong size - a lot of hotels have little interest in booking conference facilities if you're going to rent only 100 rooms.
<ijchain> <CoderX2> this hotel has a serious lack of Coke
<ijchain> <tclguy> 100 rooms isn't even right
<ijchain> <tclguy> expect more like 50 rooms
<ijchain> <tclguy> there are always local people
<azbridge> <suchenwi> Ah right - parens are not special in switch -regexp
<ijchain> <tclguy> and the few that stick to certain hotel chaings
<ijchain> <stever> this are has a wide selection
<ijchain> <stever> area
<ijchain> <rmax> ... and people who go for a cheaper hotel or motel in the neighborhood of the conference hotel.
<azbridge> * suchenwi has LP at hand :p
<ijchain> <CoderX2> i don't think we had that issue this year, this hotel is pretty cheap for this area...
<trebor_dki> oh. set t {...} instead of "..."
<ijchain> <stever> last year we had the extra wine :)
<ijchain> <stever> applause!
<trebor_dki> sorry, switch -regexp $t $m {puts toll;} default {puts nicht} works, i have to reread, sorry
<ijchain> <aku> clap clap clap
<ijchain> <stever> Thanks Reinhard!
<ijchain> <das> thx reinhard!
<ijchain> <aku> Thanks rmax for TCR
<ijchain> <schelte> Thanks Reinhard
<ijchain> <kbk> Next year in Jerusalem. Or at least Joisey.
<ijchain> <tpoindex> Danke, Reinhard
<ijchain> <rmax> Streams finally down.
<xet7> Kiiros, Reinhard & all from Finland
<xet7> kiitos
<xet7> argh typos...
<ijchain> <steveo> sp remembers ... duh
<trebor_dki> what is the difference in switch -regexp $t { $m {puts ok} defalut {puts not} } and switch -regexp $t $m {puts ok} default {puts not} ?
<ijchain> <stever> luckily Im an amature plumber
<ijchain> <stever> and dentist, welder, electrician....
<ijchain> <steveo> 2 requirements to be plumber
<ijchain> <steveo> 1. friday is payday
<ijchain> <steveo> 2 s*it doesnt run uphill
<ijchain> <schelte> In switch -regexp $t $t is the value and $m the pattern. In switch -regexp $t $m it's the other way around.
<ijchain> <suchenwi> trebor_dki - Big difference!
<ijchain> <suchenwi> The typical pattern is: switch -regexp $input {re1 {action1} re2 {action2} ... default {actionN}}
<ijchain> <suchenwi> But inside the braced block, the re* parts are not substituted (because braces prevent that)
<ijchain> <suchenwi> If you want variable substitution, better use the unbraced style:
<ijchain> <suchenwi> switch -regexp $input \
<ijchain> <suchenwi> $re1 {action1} \
<ijchain> <suchenwi> $re2 {action2} \
<ijchain> <suchenwi> default {actionN}
<ijchain> <schelte> Oh yeah, that too.
<ijchain> <suchenwi> (sep) 2 % switch -regexp abc {{a(b)c} {puts 1} default {puts 0}}
<ijchain> <suchenwi> 1
<ijchain> <suchenwi> (sep) 3 % set re {a(b)c}
<ijchain> <suchenwi> a(b)c
<ijchain> <suchenwi> (sep) 4 % switch -regexp abc {$re {puts 1} default {puts 0}}
<ijchain> <suchenwi> 0
<ijchain> <suchenwi> "$re" is taken literally - "the end of the string, and then the letters r and e"... :p
<ijchain> <suchenwi> ...which will never hit.
<trebor_dki> oh, i see. ok. it works now. i have to admit i read this in Welch, but apparently i did not understand it in the right way - now it is clear ... thanks
<ijchain> <suchenwi> welcome :D
<ijchain> <kbk> Also, with non-constant re's, it's considered better style (and is also faster) to do
<ijchain> <kbk> if {[regexp $re1 $string]} {
<ijchain> <kbk> ... stuff ...
<ijchain> <kbk> } elseif {[regexp $re2 $string]} {
<ijchain> <kbk> ... more stuff...
<ijchain> <kbk> }
<ijchain> <suchenwi> Better than [switch -regexp] ?
<ijchain> <kbk> Yeah, because of the possibility of accidents with all the \'s needed to join the lines and make it look good.
<ijchain> <suchenwi> Indeed... I always used constant regexps in switches.
<ijchain> <miguel> kbk: were you here earlier for Vitale's talk?
<ijchain> <kbk> No, alas, I just joined in the middle of the town meeting
<ijchain> <miguel> interestig bits I managed to pick up: (1) TEBC has trouble with the i-cache (not data?)
<ijchain> <miguel> (2) switch screws up the pipeline - too many mispredictions
<ijchain> <kbk> miguel - Interesting, I'd have guessed that it was a L2-cache overflow. Good to know!
<ijchain> <miguel> (3) Tcl's troubles with dispatch speed are not key, as they would be for langs with smaller bytecodes
<ijchain> <miguel> (4) got interesting improvements from little tweaks (like inlining the INST_PUSH code)
<ijchain> <kbk> My copy of his draft mostly discusses Sparc performance. Does he have insights that apply to Intel?
<ijchain> <miguel> (1) and (2) tell me that I should really get down and test that indirect-threading thing
<ijchain> <miguel> he said something about subroutine-threading, not sure if he was asm'ing it by hand or what
<ijchain> <miguel> he did say something about "even more important on P4" - but it did not sound like measurements, just going by the size of the pipeline
<ijchain> <miguel> anyway - I'm dying to get the paper
<ijchain> <miguel> but there's probably not much more than a 20-30% improvement hiding under the dispatch mechanism
<ijchain> <tpoindex> miguel: http://www.cs.toronto.edu/~bv/tcl2005/ (via google)
<ijchain> <kbk> http://www.eecg.toronto.edu/~tsa/papers/ViAb04.pdf is is IVME paper.
<ijchain> <kbk> s/is is/is his/
<ijchain> <kbk> http://www.cs.toronto.edu/~bv/tcl2005/tcl2005-slides.pdf
<ijchain> <miguel> tom: thanks! google: thanks! mom: why didn't you give me a brain?
<ijchain> <suchenwi> :p
<ijchain> <miguel> LP
<ijchain> <aku> ... printing ...
<ijchain> <suchenwi> Does Jabber have a hickup?
<ijchain> <kbk> suchenwi - I thought it was just the local firewall; I've been dropped and reconnected four times in the last 10 min or so.
<ijchain> <suchenwi> I've noticed :)
<ijchain> <suchenwi> But steveo and miguel were also reported as revolving-door users, each once
<ijchain> <miguel> I have a flaky conn
<ijchain> <aku> miguel - google for 'context threading interpreter' to find additional papers about this ...
<ijchain> <LucMove> hi, everyone. conference still going on?
<ijchain> <kbk> Luciano - No, they've all gone for |_P or to the airport.
<ijchain> <LucMove> :-(
<ijchain> <kbk> All things come to an end, eventually.
<ijchain> <LucMove> couldn't hear anything today. lots of work
<ijchain> <suchenwi> s/LP/|_P
<ijchain> <LucMove> I don't even have LP...
<ijchain> <LucMove> is patthoyts available?
<ijchain> <suchenwi> LP is our emoticon for coffee, while |_P is for beer
<ijchain> <LucMove> :-)
<ijchain> <LucMove> coffee hardly ever crosses my mind
<ijchain> <suchenwi> In Tkchat, they are rendered as according icons
<ijchain> <LucMove> yes, I remember
<ijchain> <Cameron> Update on What Is Important: http://www.therawfeed.com/2005/08/battle-brewing-between-dell-microsoft.html
Return to tcl
or
Go to some related
logs:
gipson de vanzare
beginner
xorg faster rendering
|
|