Algebra Rizing
in [Lisp]
|
From: russell_mcmanus on 2 Jul 2010 16:07 Kenneth Tilton <kentilton(a)gmail.com> writes: > If it makes you feel better, performance is indeed a mission-critical > concern in my mind and I have my eye out for problems. An initial big > concern was indeed the per-key round-trip. So far I have seen zero > problems. If someone is far away from the web server network wise, then they are going to see dramatically worse responsiveness then you. -russ
From: Kenneth Tilton on 2 Jul 2010 17:57 Peder O. Klingenberg wrote: > Kenneth Tilton <kentilton(a)gmail.com> writes: > >> What's your speakeasy.net timing to DC? > > From home: 9/4 Mbit down/up. I didn't see anywhere on speakeasy.net to > measure latency, but I know from experience that rtt to the US east > coast is usually in the 150-200 ms range. > >> Is that a 32-bit computer? Tho my old XP does fine. > > I haven't used a 32-bit computer for years. 64-bit Ubuntu at work, > 64-bit OSX at home. Though FF on the mac is a 32 bit application. Well, it sounds like you are getting one minute where I am getting 4s and our environments are exactly the same. My thinking is that we can get one minute because of some issue somewhere but we cannot get 4s by accident, not reliably and repeatedly on many browsers and many OSes, so I am not worrying about this just yet. > >>> I sometimes had trouble editing text I wrote, >>> and I couln't type < or > at all. Ah, that was definitely the old version. > I clicked the "solved" button >>> prematurely, and got an error message telling me to do more work, but I >>> was unable to return to editing the text. >> Oh, I thought I fixed that. Maybe it is in the next release. > > This was a few days ago, as I said. Maybe it was fixed in the > meantime. On second thought, I was a bit imprecise. I think I may have > been able to edit the text, but not to advance to the next step in the > process. Hmmm, I do think I pushed the new version in the past 2-3 days. The old version definitely did not let me resume work after I had mistakenly said I had reached the answer. > >> The whole application went down tools. I restarted a couple of hours agao. > > Right. It was back for a moment, but none of the choices on the > freestyle tab did anything, so I never got to editing anything. Did you try clicking "OK"? > A > page reload gave me infinite "waiting for teamalgebra.com", and further > reloads gave me server resets or the rather boring empty gray page. How recently? I restarted the server/app a few hours agao and it is still up and responding, no backtrace in sight. > >> You were not demoing Cells. > > I thought this was about hooking cells up to the js gui stuff? As I > have zero interest in the webgui stuff, I was hoping to get a glimpse of > how cells could drive an application efficiently. You need the source for RoboCells, the application that does nothing but talk to a soccer server over a socket. Pure Cells (plus football). kt -- http://www.stuckonalgebra.com "The best Algebra tutorial program I have seen... in a class by itself." Macworld
From: Kenneth Tilton on 2 Jul 2010 18:16 Kenneth Tilton wrote: > Peder O. Klingenberg wrote: >> I thought this was about hooking cells up to the js gui stuff? As I >> have zero interest in the webgui stuff, I was hoping to get a glimpse of >> how cells could drive an application efficiently. Dude, just work from first principles. Cells would let me specify: (make-instance 'thing-1 :d (the-formula (if a b c)) (if a b c) being the rule. Dependencies are decided dynamically each time a rule runs, so if a is true there is no dependency on c (in the cells world or in the real world). Thus if c changes there is no real-world reason to run this rule. If a goes to false there will be no dependency on b and vice versa. The beauty is there is no explicit declaration of dependency by the hard-working programmer -- we just write code naturally and efficient recalculation results. In the face of change, the system updates that absolute minimum of state. But! I get a kick out of the fact that this breaks that efficiency: (let ((x (based-on b)) (y (based-on c))) (if a x y)) ....because b and c are always read and reading something is used to denote dependency. ie, Somehow Cells punishes bad code (why calculate both x and y before knowing which will be needed?). When things like that emerge I know I have tapped into fundamental soundness: ie, we should have been programming like this from day one. kt -- http://teamalgebra.com http://www.stuckonalgebra.com "The best Algebra tutorial program I have seen... in a class by itself." Macworld
From: Kenneth Tilton on 2 Jul 2010 21:20 Kenneth Tilton wrote: > Peder O. Klingenberg wrote: >> Kenneth Tilton <kentilton(a)gmail.com> writes: >> >>> What's your speakeasy.net timing to DC? >> >> From home: 9/4 Mbit down/up. I didn't see anywhere on speakeasy.net to >> measure latency, but I know from experience that rtt to the US east >> coast is usually in the 150-200 ms range. >> >>> Is that a 32-bit computer? Tho my old XP does fine. >> >> I haven't used a 32-bit computer for years. 64-bit Ubuntu at work, >> 64-bit OSX at home. Though FF on the mac is a 32 bit application. > > Well, it sounds like you are getting one minute where I am getting 4s > and our environments are exactly the same. > > My thinking is that we can get one minute because of some issue > somewhere but we cannot get 4s by accident, not reliably and repeatedly > on many browsers and many OSes, so I am not worrying about this just yet. Hmmm. This would not affect the initial load because they are brought over one character at a time, but unless one has installed TeX fonts: http://www.math.union.edu/~dpvc/jsmath/download/jsMath-fonts.html ....jsMath falls back on image fonts. I /do/ have the TeX fonts installed (and I had failed to provide all the image fonts where jsMath could find them, which is why you did not see > and < when I was seeing them) so that would explain at least different typing speed. A little. As I said, jsMath brings characters over one at a time on demand. kt -- http://teamalgebra.com http://www.stuckonalgebra.com "The best Algebra tutorial program I have seen... in a class by itself." Macworld
From: Antony on 3 Jul 2010 01:51
On 7/2/2010 5:20 AM, Alessio Stalla wrote: > Imho, JS libraries for "RIAs" are complete bullshit. They basically > rewrite everything down to the layout manager, not leveraging the > browser at all... The part about layout manager. I haven't done much CSS, every time I saw the tricks needed to create a simple column layout I gave up (I had the luxury mostly cause I wasn't responsible for the UI). On one hand I have forever heard don't use tables, use divs and css, on the other hand one needs disproportionate effort doing the same thing using divs and css. And then it is usually not flexible if you wnat add a couple of more columns. Then there is span which seems to be utterly useless cause you can't put any block elements in it. People talk about content and representation separation, yet div and span are tied to visuals cause one says 'block' and allows certain DOM strcuture and the other does not. AFAIK there isn't a layout manager built into a web browser from the web developer point of view. What should one do if one needs a RIA and it has to be web based Are we saying HTML+CSS+Javascript only RIA solution is still way off Does HTML5 change anything above. Is there any popular book for managing layouts using HTML/CSS/Javascript (unlike common lisp, this web stuff is way too popular, means way too much effort to find the good stuff out of the mountain of things google brings up) -Antony |