Difference between revisions of "File talk:KNN.jar"

From Robowiki
Jump to navigation Jump to search
(→‎File format: new section)
 
(13 intermediate revisions by 3 users not shown)
Line 6: Line 6:
  
 
The current version uploaded already lists the repetition number, and well... I personally don't think it is realistic to do tests so large that a repetition takes longer than a couple minutes. 40 dimensions? That would take an extremely long time unless the number of points is fairly tiny I'm pretty sure, it probably hasn't locked up. --[[User:Rednaxela|Rednaxela]] 14:51, 11 March 2010 (UTC)
 
The current version uploaded already lists the repetition number, and well... I personally don't think it is realistic to do tests so large that a repetition takes longer than a couple minutes. 40 dimensions? That would take an extremely long time unless the number of points is fairly tiny I'm pretty sure, it probably hasn't locked up. --[[User:Rednaxela|Rednaxela]] 14:51, 11 March 2010 (UTC)
 +
 +
Normally I use:
 +
<pre>
 +
 +
System.out.printf("Running repetition %d of %d; %d%% completed.\r", i+1, numReps, (i+1)/numReps);
 +
 +
</pre>
 +
 +
So it didn't waste the screen when I run 100+ repetitions. Note that I haven't test this code yet (I'm not sure about %% in Java's printf, but that is what I have done in C). --[[User:Nat|<span style="color:#099;">Nat</span>]] [[User talk:Nat|<span style="color:#0a5;">Pavasant</span>]] 15:26, 11 March 2010 (UTC)
 +
 +
  
 
== File format ==
 
== File format ==
 +
Red, is there any problems with my build file? I am now having a lot of problems merging your change into my modified code. If you don't mind I'd like you to re-package it with my build file... --[[User:Nat|<span style="color:#099;">Nat</span>]] [[User talk:Nat|<span style="color:#0a5;">Pavasant</span>]] 14:54, 11 March 2010 (UTC)
  
Red, is there any problems with my build file? I am now having a lot of problems merging your change into my modified code. If you don't mind I'd like you to re-package it with my build file... --[[User:Nat|<span style="color:#099;">Nat</span>]] [[User talk:Nat|<span style="color:#0a5;">Pavasant</span>]] 14:54, 11 March 2010 (UTC)
+
I'd think the largish amount of refactoring I did would be more likely to cause issues merging. Anyway, I wasn't using the build file because I found it just creates a big hassle to use and slowed down the process of testing things. I could use it probably, but may I ask, why do you have it set up like that anyway? It seems awkward to have a source jar inside the binary jar in that manner. --[[User:Rednaxela|Rednaxela]] 15:03, 11 March 2010 (UTC)
 +
 
 +
Because if you are using Eclipse, the src.jar's content is exactly in Eclipse Project Format. I can guess you didn't use Eclipse then? I found that if I package it normal way, I'll have problem merging it into my workspace after I publish it. Perhaps we need central RoboWiki SVN repository or something like that... (I prefer Mercurial, though) --[[User:Nat|<span style="color:#099;">Nat</span>]] [[User talk:Nat|<span style="color:#0a5;">Pavasant</span>]] 15:14, 11 March 2010 (UTC)
 +
 
 +
Yeah, I'm not using Eclipse these days, so that doesn't work for me. How about [http://github.com/ github]? --[[User:Rednaxela|Rednaxela]] 15:29, 11 March 2010 (UTC)
 +
 
 +
: Was thinking the same. =) --[[User:Voidious|Voidious]] 15:33, 11 March 2010 (UTC)
 +
 
 +
: Because git confused me, and I'm using Windows where Git isn't really reliable. =) But I can use it if that's what all of you want. If anyone could provide me with command-comparison-chart between Git and Mercurial would be nice. I have to use Git for some projects still. (Cloning other's not-on-project-hosting source code mostly) --[[User:Nat|<span style="color:#099;">Nat</span>]] [[User talk:Nat|<span style="color:#0a5;">Pavasant</span>]] 15:57, 11 March 2010 (UTC)
 +
 
 +
:: For exactly such a chart, see [http://mercurial.selenic.com/wiki/GitConcepts#Command_equivalence_table here] (funny how many simple things in Mercurial require extension that are standard in GIT). As an aside, I've never seen a code hosting service quite as nice as simple as github before personally. I suppose Mercurial on Google Code could be decent, but projects in that are a bit more 'heavyweight'. --[[User:Rednaxela|Rednaxela]] 16:02, 11 March 2010 (UTC)
 +
 
 +
::: That chart doesn't have all commands, or I can assume all other commands are the same (clone, commit, push, etc.)? For Mercurial equivalent to GitHub, you can use [http://bitbucket.org BitBucket]... --[[User:Nat|<span style="color:#099;">Nat</span>]] [[User talk:Nat|<span style="color:#0a5;">Pavasant</span>]] 02:45, 12 March 2010 (UTC)
 +
 
 +
:::: Largely the same I believe, but really best to check the documentation. There are also some small conceptual differences. [http://rg03.wordpress.com/2009/04/07/mercurial-vs-git/ This] looks like a nice comparison. Also note the need to 'add' before 'commit' as noted on that chart. But sure, BitBucket looks good, perhaps it would be good to have a BitBucket repository for the KNNBenchmark? Should I make it? --[[User:Rednaxela|Rednaxela]] 03:12, 12 March 2010 (UTC)
 +
:::: Why not? But if you still prefer GitHub I can use it =) --[[User:Nat|<span style="color:#099;">Nat</span>]] [[User talk:Nat|<span style="color:#0a5;">Pavasant</span>]] 07:16, 12 March 2010 (UTC)
 +
::::: Got the source checked in [http://bitbucket.org/rednaxela/knn-benchmark here] now :) --[[User:Rednaxela|Rednaxela]] 17:40, 12 March 2010 (UTC)

Latest revision as of 18:40, 12 March 2010

Percentage

I don't suppose you can add something that will give an an idea about how far along it is? Since some times it takes so long for doing say, 40 dimensions, I wonder if it has locked up. --Chase 14:38, 11 March 2010 (UTC)

Well, there is a way, but I suppose it will make the whole running longer. I'll try to update it. --Nat Pavasant 14:45, 11 March 2010 (UTC)

The current version uploaded already lists the repetition number, and well... I personally don't think it is realistic to do tests so large that a repetition takes longer than a couple minutes. 40 dimensions? That would take an extremely long time unless the number of points is fairly tiny I'm pretty sure, it probably hasn't locked up. --Rednaxela 14:51, 11 March 2010 (UTC)

Normally I use:


System.out.printf("Running repetition %d of %d; %d%% completed.\r", i+1, numReps, (i+1)/numReps);

So it didn't waste the screen when I run 100+ repetitions. Note that I haven't test this code yet (I'm not sure about %% in Java's printf, but that is what I have done in C). --Nat Pavasant 15:26, 11 March 2010 (UTC)


File format

Red, is there any problems with my build file? I am now having a lot of problems merging your change into my modified code. If you don't mind I'd like you to re-package it with my build file... --Nat Pavasant 14:54, 11 March 2010 (UTC)

I'd think the largish amount of refactoring I did would be more likely to cause issues merging. Anyway, I wasn't using the build file because I found it just creates a big hassle to use and slowed down the process of testing things. I could use it probably, but may I ask, why do you have it set up like that anyway? It seems awkward to have a source jar inside the binary jar in that manner. --Rednaxela 15:03, 11 March 2010 (UTC)

Because if you are using Eclipse, the src.jar's content is exactly in Eclipse Project Format. I can guess you didn't use Eclipse then? I found that if I package it normal way, I'll have problem merging it into my workspace after I publish it. Perhaps we need central RoboWiki SVN repository or something like that... (I prefer Mercurial, though) --Nat Pavasant 15:14, 11 March 2010 (UTC)

Yeah, I'm not using Eclipse these days, so that doesn't work for me. How about github? --Rednaxela 15:29, 11 March 2010 (UTC)

Was thinking the same. =) --Voidious 15:33, 11 March 2010 (UTC)
Because git confused me, and I'm using Windows where Git isn't really reliable. =) But I can use it if that's what all of you want. If anyone could provide me with command-comparison-chart between Git and Mercurial would be nice. I have to use Git for some projects still. (Cloning other's not-on-project-hosting source code mostly) --Nat Pavasant 15:57, 11 March 2010 (UTC)
For exactly such a chart, see here (funny how many simple things in Mercurial require extension that are standard in GIT). As an aside, I've never seen a code hosting service quite as nice as simple as github before personally. I suppose Mercurial on Google Code could be decent, but projects in that are a bit more 'heavyweight'. --Rednaxela 16:02, 11 March 2010 (UTC)
That chart doesn't have all commands, or I can assume all other commands are the same (clone, commit, push, etc.)? For Mercurial equivalent to GitHub, you can use BitBucket... --Nat Pavasant 02:45, 12 March 2010 (UTC)
Largely the same I believe, but really best to check the documentation. There are also some small conceptual differences. This looks like a nice comparison. Also note the need to 'add' before 'commit' as noted on that chart. But sure, BitBucket looks good, perhaps it would be good to have a BitBucket repository for the KNNBenchmark? Should I make it? --Rednaxela 03:12, 12 March 2010 (UTC)
Why not? But if you still prefer GitHub I can use it =) --Nat Pavasant 07:16, 12 March 2010 (UTC)
Got the source checked in here now :) --Rednaxela 17:40, 12 March 2010 (UTC)