[Pharo-project] rubric performace
Hi, I did a little experiment to check the performance of Rubric when dealing with large content. And by large, I mean 45 MB or about 15 million lines of text. I opened the file in several other editors: - TextMate: 10s + the text is wrapped and can be browsed smoothly - Sublime Text 2: 10s + the text is wrapped and can be browsed smoothly - Eclipse: 5s + browsing is very slow (so it is basically useless) How did Rubric perform: - When opened with unwrapped lines: 33s + the text can be browsed smoothly - When opened with wrapped lines -> Pharo crashed :( In any case, I think being 3 times slower (even if without wrapping) than TextMate is highly exciting. Great job Alain. If you want to reproduce the experiment, here is the Smalltalk snippet I used: --- Gofer new smalltalkhubUser: 'AlainPlantec' project: 'Rubric'; package: 'Rubric-AlainPlantec.26'; load. [ contents := ((FileSystem disk workingDirectory / 'src' / 'ArgoUML-0-34' / 'output.mse') readStreamDo: #contents). (RubLipsumBasicExample windowWithScrolledText: ( RubLipsumBasicExample new string: contents; newScrolledTextNotWrapped)) openInWorld ] timeToRun --- Cheers, Doru -- www.tudorgirba.com "What we can governs what we wish."
Hi,
I did a little experiment to check the performance of Rubric when dealing with large content. And by large, I mean 45 MB or about 15 million lines of text.
I opened the file in several other editors: - TextMate: 10s + the text is wrapped and can be browsed smoothly - Sublime Text 2: 10s + the text is wrapped and can be browsed smoothly - Eclipse: 5s + browsing is very slow (so it is basically useless)
thanks doru this is really interesting.
How did Rubric perform: - When opened with unwrapped lines: 33s + the text can be browsed smoothly - When opened with wrapped lines -> Pharo crashed :(
In any case, I think being 3 times slower (even if without wrapping) than TextMate is highly exciting. Great job Alain.
I think that with more ironing and optimisation it will only get better! Thanks synectique http://www.synectique.eu for sponsoring the work of Alain and releasing it MIT.
If you want to reproduce the experiment, here is the Smalltalk snippet I used:
--- Gofer new smalltalkhubUser: 'AlainPlantec' project: 'Rubric'; package: 'Rubric-AlainPlantec.26'; load.
[ contents := ((FileSystem disk workingDirectory / 'src' / 'ArgoUML-0-34' / 'output.mse') readStreamDo: #contents). (RubLipsumBasicExample windowWithScrolledText: ( RubLipsumBasicExample new string: contents; newScrolledTextNotWrapped)) openInWorld ] timeToRun ---
Cheers, Doru
-- www.tudorgirba.com
"What we can governs what we wish."
On 25 April 2013 00:03, Tudor Girba <tudor@tudorgirba.com> wrote:
Hi,
I did a little experiment to check the performance of Rubric when dealing with large content. And by large, I mean 45 MB or about 15 million lines of text.
I opened the file in several other editors: - TextMate: 10s + the text is wrapped and can be browsed smoothly - Sublime Text 2: 10s + the text is wrapped and can be browsed smoothly - Eclipse: 5s + browsing is very slow (so it is basically useless)
How did Rubric perform: - When opened with unwrapped lines: 33s + the text can be browsed smoothly - When opened with wrapped lines -> Pharo crashed :(
is it reproducible?
In any case, I think being 3 times slower (even if without wrapping) than TextMate is highly exciting. Great job Alain.
If you want to reproduce the experiment, here is the Smalltalk snippet I used:
--- Gofer new smalltalkhubUser: 'AlainPlantec' project: 'Rubric'; package: 'Rubric-AlainPlantec.26'; load.
[ contents := ((FileSystem disk workingDirectory / 'src' / 'ArgoUML-0-34' / 'output.mse') readStreamDo: #contents). (RubLipsumBasicExample windowWithScrolledText: ( RubLipsumBasicExample new string: contents; newScrolledTextNotWrapped)) openInWorld ] timeToRun ---
Cheers, Doru
-- www.tudorgirba.com
"What we can governs what we wish."
-- Best regards, Igor Stasenko.
Thanx for benchmarking Rubric. Alain, is doing great work and I'll integrate in Glamour for a demo. usman On Thu, Apr 25, 2013 at 12:03 AM, Tudor Girba <tudor@tudorgirba.com> wrote:
Hi,
I did a little experiment to check the performance of Rubric when dealing with large content. And by large, I mean 45 MB or about 15 million lines of text.
I opened the file in several other editors: - TextMate: 10s + the text is wrapped and can be browsed smoothly - Sublime Text 2: 10s + the text is wrapped and can be browsed smoothly - Eclipse: 5s + browsing is very slow (so it is basically useless)
How did Rubric perform: - When opened with unwrapped lines: 33s + the text can be browsed smoothly - When opened with wrapped lines -> Pharo crashed :(
In any case, I think being 3 times slower (even if without wrapping) than TextMate is highly exciting. Great job Alain.
If you want to reproduce the experiment, here is the Smalltalk snippet I used:
--- Gofer new smalltalkhubUser: 'AlainPlantec' project: 'Rubric'; package: 'Rubric-AlainPlantec.26'; load.
[ contents := ((FileSystem disk workingDirectory / 'src' / 'ArgoUML-0-34' / 'output.mse') readStreamDo: #contents). (RubLipsumBasicExample windowWithScrolledText: ( RubLipsumBasicExample new string: contents; newScrolledTextNotWrapped)) openInWorld ] timeToRun ---
Cheers, Doru
-- www.tudorgirba.com
"What we can governs what we wish."
doru can you reproduce your crash? Because igor asked and I did not see your answer. Stef On Apr 25, 2013, at 12:03 AM, Tudor Girba <tudor@tudorgirba.com> wrote:
Hi,
I did a little experiment to check the performance of Rubric when dealing with large content. And by large, I mean 45 MB or about 15 million lines of text.
I opened the file in several other editors: - TextMate: 10s + the text is wrapped and can be browsed smoothly - Sublime Text 2: 10s + the text is wrapped and can be browsed smoothly - Eclipse: 5s + browsing is very slow (so it is basically useless)
How did Rubric perform: - When opened with unwrapped lines: 33s + the text can be browsed smoothly - When opened with wrapped lines -> Pharo crashed :(
In any case, I think being 3 times slower (even if without wrapping) than TextMate is highly exciting. Great job Alain.
If you want to reproduce the experiment, here is the Smalltalk snippet I used:
--- Gofer new smalltalkhubUser: 'AlainPlantec' project: 'Rubric'; package: 'Rubric-AlainPlantec.26'; load.
[ contents := ((FileSystem disk workingDirectory / 'src' / 'ArgoUML-0-34' / 'output.mse') readStreamDo: #contents). (RubLipsumBasicExample windowWithScrolledText: ( RubLipsumBasicExample new string: contents; newScrolledTextNotWrapped)) openInWorld ] timeToRun ---
Cheers, Doru
-- www.tudorgirba.com
"What we can governs what we wish."
Hi, I tried but could not :(. I tried multiple times to load the file and go around the "out of memory" limitation and then at some random point, Pharo crashed. I was on Mac when I did the experiment. If I get more info I will let you know. Doru On Apr 29, 2013, at 8:44 PM, stephane ducasse <stephane.ducasse@free.fr> wrote:
doru can you reproduce your crash? Because igor asked and I did not see your answer.
Stef
On Apr 25, 2013, at 12:03 AM, Tudor Girba <tudor@tudorgirba.com> wrote:
Hi,
I did a little experiment to check the performance of Rubric when dealing with large content. And by large, I mean 45 MB or about 15 million lines of text.
I opened the file in several other editors: - TextMate: 10s + the text is wrapped and can be browsed smoothly - Sublime Text 2: 10s + the text is wrapped and can be browsed smoothly - Eclipse: 5s + browsing is very slow (so it is basically useless)
How did Rubric perform: - When opened with unwrapped lines: 33s + the text can be browsed smoothly - When opened with wrapped lines -> Pharo crashed :(
In any case, I think being 3 times slower (even if without wrapping) than TextMate is highly exciting. Great job Alain.
If you want to reproduce the experiment, here is the Smalltalk snippet I used:
--- Gofer new smalltalkhubUser: 'AlainPlantec' project: 'Rubric'; package: 'Rubric-AlainPlantec.26'; load.
[ contents := ((FileSystem disk workingDirectory / 'src' / 'ArgoUML-0-34' / 'output.mse') readStreamDo: #contents). (RubLipsumBasicExample windowWithScrolledText: ( RubLipsumBasicExample new string: contents; newScrolledTextNotWrapped)) openInWorld ] timeToRun ---
Cheers, Doru
-- www.tudorgirba.com
"What we can governs what we wish."
-- www.tudorgirba.com "From an abstract enough point of view, any two things are similar."
tx! On Apr 29, 2013, at 9:24 PM, Tudor Girba <tudor@tudorgirba.com> wrote:
Hi,
I tried but could not :(. I tried multiple times to load the file and go around the "out of memory" limitation and then at some random point, Pharo crashed. I was on Mac when I did the experiment. If I get more info I will let you know.
Doru
On Apr 29, 2013, at 8:44 PM, stephane ducasse <stephane.ducasse@free.fr> wrote:
doru can you reproduce your crash? Because igor asked and I did not see your answer.
Stef
On Apr 25, 2013, at 12:03 AM, Tudor Girba <tudor@tudorgirba.com> wrote:
Hi,
I did a little experiment to check the performance of Rubric when dealing with large content. And by large, I mean 45 MB or about 15 million lines of text.
I opened the file in several other editors: - TextMate: 10s + the text is wrapped and can be browsed smoothly - Sublime Text 2: 10s + the text is wrapped and can be browsed smoothly - Eclipse: 5s + browsing is very slow (so it is basically useless)
How did Rubric perform: - When opened with unwrapped lines: 33s + the text can be browsed smoothly - When opened with wrapped lines -> Pharo crashed :(
In any case, I think being 3 times slower (even if without wrapping) than TextMate is highly exciting. Great job Alain.
If you want to reproduce the experiment, here is the Smalltalk snippet I used:
--- Gofer new smalltalkhubUser: 'AlainPlantec' project: 'Rubric'; package: 'Rubric-AlainPlantec.26'; load.
[ contents := ((FileSystem disk workingDirectory / 'src' / 'ArgoUML-0-34' / 'output.mse') readStreamDo: #contents). (RubLipsumBasicExample windowWithScrolledText: ( RubLipsumBasicExample new string: contents; newScrolledTextNotWrapped)) openInWorld ] timeToRun ---
Cheers, Doru
-- www.tudorgirba.com
"What we can governs what we wish."
-- www.tudorgirba.com
"From an abstract enough point of view, any two things are similar."
participants (5)
-
Igor Stasenko -
stephane ducasse -
Stéphane Ducasse -
Tudor Girba -
Usman Bhatti