[Moose-dev] Re: can you say to alex that edges are slow?

Hi, > yes but there is something that I don't understand: > why each node and edges does not keep these state ? > width and height are cool to keep, but this is the same problem for > color, border.... > Each of them could have computation beside; so the problem is in > general. > > Maybe my vision is naive but if each node and edge keep its state, > we do not need a cache or multiple computations. This is the correct point of view. The invalidation of the visualization should be controlled by the developer. Doru > There is something not clear for me :) > > On Jun 9, 2010, at 22:19 , Alexandre Bergel wrote: > >> I can add something like: >> MOShape>>enableCache >> MOShape>>disableCache >> >> In that case, width: and height: will be computed just once per >> node. >> >> Cheers, >> Alexandre >> >> On 9 Jun 2010, at 16:09, Laval Jannik wrote: >> >>> Hi Doru, >>> >>> Sure, we will see the computation in blueprint, and make a cache. >>> Now, the problem is more general: >>> I have the same problem with eDSM: Some edges and nodes make >>> computation during rendering, which slow down the visualization. >>> >>> So, my question is more general: maybe Mondrian should integrate >>> the cache system, not each work we do. >>> >>> Cheers, >>> Jannik >>> >>> >>> On Jun 9, 2010, at 21:58 , Tudor Girba wrote: >>> >>>> Hi Jannik, >>>> >>>> On 9 Jun 2010, at 17:23, Laval Jannik wrote: >>>> >>>>> Hi all, >>>>> >>>>> Arf, the transmission of the file has probably failed. >>>>> So, what is done with bluprint: >>>>> >>>>> Load Moose in Moose, >>>>> Take all model packages, generate outgoing blueprint. >>>>> It is easier like that :) >>>>> >>>>> But, there is something that I do not understand: why Mondrian >>>>> compute all the time. >>>>> Should not it compute one time and put in cache results ? >>>>> If it is done in Mondrian, it will be more compatible with all >>>>> things that we can do ? >>>> >>>> I am not saying that Mondrian is doing that, I am saying that if >>>> we want to debug the problem we want to isolate the case :). >>>> That is why, if we just want to see how slow/fast the rendering >>>> is, we want as less extra computation as possible in the >>>> benchmark. >>>> >>>> >>>> Cheers, >>>> Doru >>>> >>>> >>>>> Cheers, >>>>> Jannik >>>>> >>>>> On Jun 9, 2010, at 17:06 , Tudor Girba wrote: >>>>> >>>>>> Hi Jannik, >>>>>> >>>>>> Something seems to be wrong with the zip file because the >>>>>> decompression fails. >>>>>> >>>>>> If you want to try just the perfromance, please just create a >>>>>> new script that explicitly creates the nodes and edges and >>>>>> sets the size and color of shapes. It could be that the >>>>>> computation is problematic, or it could be that Mondrian calls >>>>>> this computation too many times. Using such a script allows us >>>>>> to get an idea of where the problem is. >>>>>> >>>>>> Cheers, >>>>>> Doru >>>>>> >>>>>> >>>>>> On 9 Jun 2010, at 16:23, Laval Jannik wrote: >>>>>> >>>>>>> Ok, I understand. >>>>>>> >>>>>>> So the perfect thing is to compute all metrics before the >>>>>>> script, or use a moose cache. >>>>>>> I think that blueprint use cook, with a cache. >>>>>>> >>>>>>> You can fing t image here: >>>>>>> http://jannik-laval.eu/assets/files/blueprint.zip >>>>>>> >>>>>>> Cheers, >>>>>>> Jannik >>>>>>> >>>>>>> >>>>>>> On Jun 9, 2010, at 16:08 , Alexandre Bergel wrote: >>>>>>> >>>>>>>>> Yes, we will do this. >>>>>>>>> Now, I do not understand the relation between metrics >>>>>>>>> (which is computed before visualization) and scrolling the >>>>>>>>> canvas. >>>>>>>>> >>>>>>>>> Does Mondrian compute something when we scroll ? >>>>>>>> >>>>>>>> Mondrian itself no. But it evaluates the block you provide >>>>>>>> when defining shape (e.g., height: and width:) each time a >>>>>>>> refresh is done. Which occurs when you scroll. >>>>>>>> I would like to verify if the metrics you use are really >>>>>>>> computed just once. >>>>>>>> >>>>>>>> Alexandre >>>>>>>> >>>>>>>>> >>>>>>>>> Cheers, >>>>>>>>> Jannik >>>>>>>>> >>>>>>>>> On Jun 9, 2010, at 15:46 , Alexandre Bergel wrote: >>>>>>>>> >>>>>>>>>> It absolutely possible that I missed opportunities for >>>>>>>>>> optimization. But I would like to be sure that the metrics >>>>>>>>>> you use are not responsible for the slowness. Can you >>>>>>>>>> produce a complete sequence on how I can reproduce the >>>>>>>>>> problem, starting from a Moose one-click image? >>>>>>>>>> >>>>>>>>>> Cheers, >>>>>>>>>> Alexandre >>>>>>>>>> >>>>>>>>>> On 9 Jun 2010, at 09:41, Laval Jannik wrote: >>>>>>>>>> >>>>>>>>>>> Hi Alex, >>>>>>>>>>> >>>>>>>>>>> I just see Package Blueprint working on Pharo1.1 >>>>>>>>>>> Jean Remi shows a case study : >>>>>>>>>>> It is a visualization of 50 packages (so, not a lot), it >>>>>>>>>>> is clearly difficult to scroll. >>>>>>>>>>> This is not an edges problem, I do not know if this >>>>>>>>>>> problem comes from Mondrian or Morphic... >>>>>>>>>>> >>>>>>>>>>> But for sure, it is not attractive. >>>>>>>>>>> >>>>>>>>>>> Mondrian exists to generate faster visualization by >>>>>>>>>>> scripting. >>>>>>>>>>> If the visualization is not scrollable, and not usable >>>>>>>>>>> with modern computers as we have, do you think Mondrian >>>>>>>>>>> is useful ? >>>>>>>>>>> >>>>>>>>>>> The problem is here: We write one time the script, we >>>>>>>>>>> generate frequently visualization. >>>>>>>>>>> Users prefer a fast visualization than a fast scripting >>>>>>>>>>> language. >>>>>>>>>>> The perfect is to have the two. And I think that Mondrian >>>>>>>>>>> can do this. >>>>>>>>>>> >>>>>>>>>>> Maybe we should do work on it (an engineer ?). >>>>>>>>>>> >>>>>>>>>>> Just my reflexion >>>>>>>>>>> >>>>>>>>>>> Cheers, >>>>>>>>>>> Jannik >>>>>>>>>>> >>>>>>>>>>> On Jun 9, 2010, at 14:24 , Alexandre Bergel wrote: >>>>>>>>>>> >>>>>>>>>>>>> I already tried with a complex case. Indeed it is a >>>>>>>>>>>>> lot faster than before, thank you :D >>>>>>>>>>>> >>>>>>>>>>>> This is what I would expect. There was a bug introduced >>>>>>>>>>>> with the zOrder. >>>>>>>>>>>> You should have similar speed up on pharo 1.0. >>>>>>>>>>>> >>>>>>>>>>>>> The #5 is a treeLayout (with less edges) and on it I >>>>>>>>>>>>> couldn't move the scrollbar.. got stuck for minutes. I >>>>>>>>>>>>> tried again, and I had the same problem. >>>>>>>>>>>> >>>>>>>>>>>> Are you sure that the problem comes from Mondrian? You >>>>>>>>>>>> can use Kai, the profiler I worked on to easily see >>>>>>>>>>>> where the problem comes from. >>>>>>>>>>>> >>>>>>>>>>>> KaiProfiler viewProfiling: [ "your code here" ] >>>>>>>>>>>> >>>>>>>>>>>> www.squeaksource.com/Spy.html >>>>>>>>>>>> >>>>>>>>>>>> Cheers, >>>>>>>>>>>> Alexandre >>>>>>>>>>>> >>>>>>>>>>>>> >>>>>>>>>>>>> Will check if i am doing something different for that >>>>>>>>>>>>> visualization (thought is the simplest one), and will >>>>>>>>>>>>> let you know. >>>>>>>>>>>>> >>>>>>>>>>>>> Regards, >>>>>>>>>>>>> Veronica >>>>>>>>>>>>> >>>>>>>>>>>>> >>>>>>>>>>>>> On 08 Jun 2010, at 14:46, Alexandre Bergel wrote: >>>>>>>>>>>>> >>>>>>>>>>>>>> Hi Veronica (and Jannik), >>>>>>>>>>>>>> >>>>>>>>>>>>>> The problem should now be fixed. The last version of >>>>>>>>>>>>>> Mondrian is 3 times faster when displaying inner nodes >>>>>>>>>>>>>> and edges. >>>>>>>>>>>>>> Does it solve your problem? >>>>>>>>>>>>>> >>>>>>>>>>>>>> Cheers, >>>>>>>>>>>>>> Alexandre >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> On 5 Jun 2010, at 05:27, Veronica Isabel Uquillas >>>>>>>>>>>>>> Gomez wrote: >>>>>>>>>>>>>> >>>>>>>>>>>>>>> Hi Alexandre, >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> Inspecting visualizations when many edges are present >>>>>>>>>>>>>>> is really a problem... >>>>>>>>>>>>>>> Is there any improvement with respect to this problem? >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> regards, >>>>>>>>>>>>>>> Veronica >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> On 24 Apr 2010, at 22:53, Alexandre Bergel wrote: >>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> Ok. I have a set of benchmarks that noticed a >>>>>>>>>>>>>>>> slowdown. I hope this will be addressed soon (a >>>>>>>>>>>>>>>> student is currently working on it). >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> Alexandre >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> On 24 Apr 2010, at 07:59, Laval Jannik wrote: >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>> I can't too. >>>>>>>>>>>>>>>>> The script make a while and crash the image. >>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>> Jannik >>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>> On Apr 23, 2010, at 23:07 , Alexandre Bergel wrote: >>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>> Jannik, can you try to generate a DSM with the >>>>>>>>>>>>>>>>>> version Mondrian-Alexandre_Bergel.355.mcz and tell >>>>>>>>>>>>>>>>>> me whether it is faster or not ? >>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>> I guess Veronica can not do it since dashed lines >>>>>>>>>>>>>>>>>> did not exist in 355. >>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>> Cheers, >>>>>>>>>>>>>>>>>> Alexandre >>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>> On 23 Apr 2010, at 15:41, Laval Jannik wrote: >>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>> Hi, >>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>> This is the same for me. >>>>>>>>>>>>>>>>>>> With an eDSM, it is really difficult to scroll. >>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>> I already make some reports of this problem... >>>>>>>>>>>>>>>>>>> But I am not sure it is caused by edges. >>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>> Jannik >>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>> On Apr 23, 2010, at 21:31 , Veronica Isabel >>>>>>>>>>>>>>>>>>> Uquillas Gomez wrote: >>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>> Hi Alexandre, >>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>> when my visualizations have many elements in >>>>>>>>>>>>>>>>>>>> general is slow, but sometimes is impossible (or >>>>>>>>>>>>>>>>>>>> takes more than 1 min) to move the scrollbar or >>>>>>>>>>>>>>>>>>>> switch to other windows... I think the edges are >>>>>>>>>>>>>>>>>>>> causing this problem, but i am not sure. >>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>> will send you an example of this case. >>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>> regards, >>>>>>>>>>>>>>>>>>>> Veronica >>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>> On 23 Apr 2010, at 20:09, Alexandre Bergel wrote: >>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>> Are edges slower than before? I recently >>>>>>>>>>>>>>>>>>>>> noticed a slowdown, but it is not that apparent. >>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>> How many edges are intended to be defined? >>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>> Cheers, >>>>>>>>>>>>>>>>>>>>> Alexandre >>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>> On 23 Apr 2010, at 13:41, stephane ducasse wrote: >>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>>> if you do not tell him, he will never notice it. >>>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>>> First Rule: give feedback! >>>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>>> Stef >>>>>>>>>>>>>>>>>>>>>> _______________________________________________ >>>>>>>>>>>>>>>>>>>>>> Moose-dev mailing list >>>>>>>>>>>>>>>>>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>>>>>>>>>>>>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>> -- >>>>>>>>>>>>>>>>>>>>> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: >>>>>>>>>>>>>>>>>>>>> Alexandre Bergel http://www.bergel.eu >>>>>>>>>>>>>>>>>>>>> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. >>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>>> _______________________________________________ >>>>>>>>>>>>>>>>>>>> Moose-dev mailing list >>>>>>>>>>>>>>>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>>>>>>>>>>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose- >>>>>>>>>>>>>>>>>>>> dev >>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>> --- >>>>>>>>>>>>>>>>>>> Jannik Laval >>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>>> _______________________________________________ >>>>>>>>>>>>>>>>>>> Moose-dev mailing list >>>>>>>>>>>>>>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>>>>>>>>>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>> -- >>>>>>>>>>>>>>>>>> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: >>>>>>>>>>>>>>>>>> Alexandre Bergel http://www.bergel.eu >>>>>>>>>>>>>>>>>> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. >>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>>> _______________________________________________ >>>>>>>>>>>>>>>>>> Moose-dev mailing list >>>>>>>>>>>>>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>>>>>>>>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>> --- >>>>>>>>>>>>>>>>> Jannik Laval >>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>>> _______________________________________________ >>>>>>>>>>>>>>>>> Moose-dev mailing list >>>>>>>>>>>>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>>>>>>>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> -- >>>>>>>>>>>>>>>> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;: >>>>>>>>>>>>>>>> Alexandre Bergel http://www.bergel.eu >>>>>>>>>>>>>>>> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;. >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> >>>>>>>>>>>>>>>> _______________________________________________ >>>>>>>>>>>>>>>> Moose-dev mailing list >>>>>>>>>>>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>>>>>>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> >>>>>>>>>>>>>>> _______________________________________________ >>>>>>>>>>>>>>> Moose-dev mailing list >>>>>>>>>>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>>>>>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>>>>>>>>>> >>>>>>>>>>>>>> >>>>>>>>>>>>>> _______________________________________________ >>>>>>>>>>>>>> Moose-dev mailing list >>>>>>>>>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>>>>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>>>>>>>>> >>>>>>>>>>>>> >>>>>>>>>>>>> _______________________________________________ >>>>>>>>>>>>> Moose-dev mailing list >>>>>>>>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>>>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>>>>>>>> >>>>>>>>>>>> >>>>>>>>>>>> _______________________________________________ >>>>>>>>>>>> Moose-dev mailing list >>>>>>>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> _______________________________________________ >>>>>>>>>>> Moose-dev mailing list >>>>>>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> _______________________________________________ >>>>>>>>>> Moose-dev mailing list >>>>>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>>>>> >>>>>>>>> --- >>>>>>>>> Jannik Laval >>>>>>>>> >>>>>>>>> >>>>>>>>> _______________________________________________ >>>>>>>>> Moose-dev mailing list >>>>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>>>> >>>>>>>> >>>>>>>> _______________________________________________ >>>>>>>> Moose-dev mailing list >>>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>>> >>>>>>> --- >>>>>>> Jannik Laval >>>>>>> >>>>>>> >>>>>>> _______________________________________________ >>>>>>> Moose-dev mailing list >>>>>>> Moose-dev(a)iam.unibe.ch >>>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>>> >>>>>> -- >>>>>> www.tudorgirba.com >>>>>> >>>>>> "Every thing should have the right to be different." >>>>>> >>>>>> >>>>>> >>>>>> _______________________________________________ >>>>>> Moose-dev mailing list >>>>>> Moose-dev(a)iam.unibe.ch >>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>>> >>>>> --- >>>>> Jannik Laval >>>>> >>>>> >>>>> _______________________________________________ >>>>> Moose-dev mailing list >>>>> Moose-dev(a)iam.unibe.ch >>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>>> >>>> -- >>>> www.tudorgirba.com >>>> >>>> "Every now and then stop and ask yourself if the war you're >>>> fighting is the right one." >>>> >>>> >>>> >>>> _______________________________________________ >>>> Moose-dev mailing list >>>> Moose-dev(a)iam.unibe.ch >>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >>> >>> --- >>> Jannik Laval >>> >>> >>> _______________________________________________ >>> Moose-dev mailing list >>> Moose-dev(a)iam.unibe.ch >>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev >> >> >> _______________________________________________ >> Moose-dev mailing list >> Moose-dev(a)iam.unibe.ch >> https://www.iam.unibe.ch/mailman/listinfo/moose-dev > > --- > Jannik Laval > > > _______________________________________________ > Moose-dev mailing list > Moose-dev(a)iam.unibe.ch > https://www.iam.unibe.ch/mailman/listinfo/moose-dev -- www.tudorgirba.com "It's not how it is, it is how we see it."