My favorites | Sign in
Project Logo
                
Details: Show all Hide all

Last 30 days

  • Nov 27, 2009
    issue 17 (Python for Ubuntu fails) Status changed by francois.parmentier   -   Can you try running it from a normal shell (python Ector.py)? Until now, I did not even try to run it from IDLE (I just checked: it does not work from IDLE, you're right). I use Eclipse to develop, so I did not note that problem.
    Status: Accepted
    Can you try running it from a normal shell (python Ector.py)? Until now, I did not even try to run it from IDLE (I just checked: it does not work from IDLE, you're right). I use Eclipse to develop, so I did not note that problem.
    Status: Accepted
  • Nov 24, 2009
    issue 17 (Python for Ubuntu fails) reported by Avaera8820   -   What steps will reproduce the problem? Attempt to run Ector using the IDLE (Python 2.6) What is the expected output? What do you see instead? Expected Ector to start. Did not start. What version of pyECTOR are you using? Latest What version of Python are you using ? 2.6, bundled in IDLE On what operating system? Ubuntu 9.10 (Karmic Koala) Please provide any additional information below. The shell returned this: Traceback (most recent call last): File "/home/ccheng/Desktop/pyector/src/Ector.py", line 721, in <module> status = main() File "/home/ccheng/Desktop/pyector/src/Ector.py", line 571, in main if stdin.closed: AttributeError: 'PyShell' object has no attribute 'closed'
    What steps will reproduce the problem? Attempt to run Ector using the IDLE (Python 2.6) What is the expected output? What do you see instead? Expected Ector to start. Did not start. What version of pyECTOR are you using? Latest What version of Python are you using ? 2.6, bundled in IDLE On what operating system? Ubuntu 9.10 (Karmic Koala) Please provide any additional information below. The shell returned this: Traceback (most recent call last): File "/home/ccheng/Desktop/pyector/src/Ector.py", line 721, in <module> status = main() File "/home/ccheng/Desktop/pyector/src/Ector.py", line 571, in main if stdin.closed: AttributeError: 'PyShell' object has no attribute 'closed'

Earlier this year

  • Oct 29, 2009
    Resources (Resources list.) Wiki page edited by francois.parmentier   -   Revision r205 Edited wiki page through web user interface.
    Revision r205 Edited wiki page through web user interface.
  • Jul 08, 2009
    Resources (Resources list.) Wiki page edited by francois.parmentier
  • May 20, 2009
    issue 15 (pyEctor think time increases with every new sentence) commented on by francois.parmentier   -   True. That has always been the trouble with ECTOR: the more nodes there are, the more time the reply will take. I already did optimize some things (in the initial PHP/MySQL version), but there is no miracle. I can't compare reply times any more, as my Quad Core is much quicker. It depends also on free memory, I think. I initially wanted to use Google App Engine, and thus to use their persistence mechanism, but this is not for today! I prefer ending the development of expression parsing, to keep a clean and readable algorithm the longest time. And there is another trouble: I do not currently spend much time on pyECTOR (but your feedback is a good motivation). (please pardon my poor english)
    True. That has always been the trouble with ECTOR: the more nodes there are, the more time the reply will take. I already did optimize some things (in the initial PHP/MySQL version), but there is no miracle. I can't compare reply times any more, as my Quad Core is much quicker. It depends also on free memory, I think. I initially wanted to use Google App Engine, and thus to use their persistence mechanism, but this is not for today! I prefer ending the development of expression parsing, to keep a clean and readable algorithm the longest time. And there is another trouble: I do not currently spend much time on pyECTOR (but your feedback is a good motivation). (please pardon my poor english)
  • May 20, 2009
    issue 16 (Answer without learning) commented on by francois.parmentier   -   I agree! I thought about that, for example to be able to integrate the bot in a game, where the "personality" of the character must not change (learning from gamers).
    I agree! I thought about that, for example to be able to integrate the bot in a game, where the "personality" of the character must not change (learning from gamers).
  • May 10, 2009
    issue 16 (Answer without learning) reported by garagoth   -   Add option for Ector to just answer to user input, without adding his sentence into knowledge base. This would make bot usable in environments where only some users are authorized to learn bot as others will try to feed junk data into knowledge base.
    Add option for Ector to just answer to user input, without adding his sentence into knowledge base. This would make bot usable in environments where only some users are authorized to learn bot as others will try to feed junk data into knowledge base.
  • May 10, 2009
    issue 15 (pyEctor think time increases with every new sentence) reported by garagoth   -   What steps will reproduce the problem? 1. Enter like 30-40 sentences, chatting with pyEctor 2. Observe response time and CPU usage What is the expected output? What do you see instead? Almost instant reply. That is, no more that 1 second to be reasonable. I wait, with every new sentence, 10 second, 20 seconds and more... @cleanstate does not help. What version of pyECTOR are you using? pyECTOR version 0.4 What version of Python are you using ? Python 2.5.2 On what operating system? Linux 2.6.22.9 #2 SMP Pentium II 350
    What steps will reproduce the problem? 1. Enter like 30-40 sentences, chatting with pyEctor 2. Observe response time and CPU usage What is the expected output? What do you see instead? Almost instant reply. That is, no more that 1 second to be reasonable. I wait, with every new sentence, 10 second, 20 seconds and more... @cleanstate does not help. What version of pyECTOR are you using? pyECTOR version 0.4 What version of Python are you using ? Python 2.5.2 On what operating system? Linux 2.6.22.9 #2 SMP Pentium II 350
  • May 10, 2009
    issue 14 (maximum recursion depth exceeded when trying to save) commented on by garagoth   -   use cPickle as pickle This line fixed my @write problems, seems to handle circular refs much better. Regards, Garagoth.
    use cPickle as pickle This line fixed my @write problems, seems to handle circular refs much better. Regards, Garagoth.
  • May 10, 2009
    issue 14 (maximum recursion depth exceeded when trying to save) commented on by k.durczak   -   use cPickle as pickle This line fixed my problems with save. Seems to handle circular refs much better. Regards, Garagoth.
    use cPickle as pickle This line fixed my problems with save. Seems to handle circular refs much better. Regards, Garagoth.
  • Apr 20, 2009
    Resources (Resources list.) Wiki page edited by francois.parmentier
  • Apr 13, 2009
    issue 14 (maximum recursion depth exceeded when trying to save) commented on by bogdan.i.ivanov   -   In the sys module, there are the methods getrecursionlimit and setrecursionlimit On my system (Acer Aspire 5520) sys.getrecursionlimit() returns 1000 I managed to sys.setrecursionlimit(200000000), but even with this setting, pickle eventually crashed. A solution to this might be to implement a custom save/load, instead of using pickle. I will try and see if I can help with this. Also, great program and concept.
    In the sys module, there are the methods getrecursionlimit and setrecursionlimit On my system (Acer Aspire 5520) sys.getrecursionlimit() returns 1000 I managed to sys.setrecursionlimit(200000000), but even with this setting, pickle eventually crashed. A solution to this might be to implement a custom save/load, instead of using pickle. I will try and see if I can help with this. Also, great program and concept.
  • Apr 08, 2009
    Resources (Resources list.) Wiki page edited by francois.parmentier
  • Apr 04, 2009
    issue 14 (maximum recursion depth exceeded when trying to save) commented on by francois.parmentier   -   As the model of ECTOR is a network, there are many circular references, and I think the maximum recursion depth is exceeded (what? that's what the error message says? true). Now, I have to look for a solution... As I have not explored this way, that article is a good beginning to me: http://www.developertutorials.com/tutorials/python/python-persistence-management-050405/page1.html
    As the model of ECTOR is a network, there are many circular references, and I think the maximum recursion depth is exceeded (what? that's what the error message says? true). Now, I have to look for a solution... As I have not explored this way, that article is a good beginning to me: http://www.developertutorials.com/tutorials/python/python-persistence-management-050405/page1.html
  • Mar 31, 2009
    Resources (Resources list.) Wiki page edited by francois.parmentier
  • Mar 18, 2009
    Resources (Resources list.) Wiki page edited by francois.parmentier
  • Jan 27, 2009
    issue 14 (maximum recursion depth exceeded when trying to save) Labels changed by francois.parmentier   -  
    Labels: Milestone-Release0.5 Milestone-Release0.4
    Labels: Milestone-Release0.5 Milestone-Release0.4
  • Jan 26, 2009
    issue 14 (maximum recursion depth exceeded when trying to save) Labels changed by francois.parmentier   -  
    Labels: Milestone-Release0.4
    Labels: Milestone-Release0.4
  • Jan 26, 2009
    issue 14 (maximum recursion depth exceeded when trying to save) commented on by francois.parmentier   -   That's true, I did not elaborate a lot on save. Can you send the log file? -- PS: I'm French, if you prefer, you can gmail me ;)
    That's true, I did not elaborate a lot on save. Can you send the log file? -- PS: I'm French, if you prefer, you can gmail me ;)
  • Jan 26, 2009
    issue 14 (maximum recursion depth exceeded when trying to save) reported by nicolas.coevoet   -   What steps will reproduce the problem? 1. stuff a lot with the bot, like listen an irc channel 2. do ector.write() What is the expected output? What do you see instead? self.memoize(obj) RuntimeError: maximum recursion depth exceeded What version of pyECTOR are you using? 0.4 Released What version of Python are you using ? On what operating system? 2.5 on debian lenny Please provide any additional information below. Good bot in fact, but this is big problem if it s impossible to save brain to disk :/
    What steps will reproduce the problem? 1. stuff a lot with the bot, like listen an irc channel 2. do ector.write() What is the expected output? What do you see instead? self.memoize(obj) RuntimeError: maximum recursion depth exceeded What version of pyECTOR are you using? 0.4 Released What version of Python are you using ? On what operating system? 2.5 on debian lenny Please provide any additional information below. Good bot in fact, but this is big problem if it s impossible to save brain to disk :/

Older

  • Dec 18, 2008
    EctorManual (Manual of the Ector program.) Wiki page edited by francois.parmentier
  • Dec 11, 2008
    pyector-0.4.tar.gz (Release 0.4 - Sentence generation working) file uploaded by francois.parmentier   -  
    Labels: Featured OpSys-All Type-Archive
    Labels: Featured OpSys-All Type-Archive
  • Dec 11, 2008
    r198 (Release 0.4 - Sentence generation working) committed by francois.parmentier   -   Release 0.4 - Sentence generation working
    Release 0.4 - Sentence generation working
  • Dec 11, 2008
    r197 (Prepare version 0.4) committed by francois.parmentier   -   Prepare version 0.4
    Prepare version 0.4
  • Dec 10, 2008
    issue 11 (Prevent infinite loop on generating sentence) changed by francois.parmentier   -   r196 fixes this.
    Status: Fixed
    Labels: Priority-Medium
    r196 fixes this.
    Status: Fixed
    Labels: Priority-Medium
  • Dec 10, 2008
    r196 (Fix issue 11. Prevent repetition in generated phrase. Di...) committed by francois.parmentier   -   Fix issue 11 . Prevent repetition in generated phrase. Divide the choice weight of each node by the number of times it already appears in the generated phrase (multiplied by the length of the symbol: short symbols may be repeated more).
    Fix issue 11 . Prevent repetition in generated phrase. Divide the choice weight of each node by the number of times it already appears in the generated phrase (multiplied by the length of the symbol: short symbols may be repeated more).
  • Dec 09, 2008
    issue 13 (Integrate Ector flow into the Ector class) reported by francois.parmentier   -   Almost all of Ector's flow is in the main() function of Ector.py, instead of being integrated to Ector's class. This would make Ector class much easier to unit test (replacing sys.stdin with a StringIO...).
    Almost all of Ector's flow is in the main() function of Ector.py, instead of being integrated to Ector's class. This would make Ector class much easier to unit test (replacing sys.stdin with a StringIO...).
  • Dec 09, 2008
    issue 12 (Make License a module to import) Status changed by francois.parmentier   -   r195 fixes this, in a more elegant manner. See http://diveintopython.org/functional_programming/finding_the_path.html
    Status: Fixed
    r195 fixes this, in a more elegant manner. See http://diveintopython.org/functional_programming/finding_the_path.html
    Status: Fixed
  • Dec 09, 2008
    r195 (Fix issue 12: the license file path is computed from the scr...) committed by francois.parmentier   -   Fix issue 12 : the license file path is computed from the script path. See http://diveintopython.org/functional_programming/finding_the_path.html
    Fix issue 12 : the license file path is computed from the script path. See http://diveintopython.org/functional_programming/finding_the_path.html
  • Dec 05, 2008
    Resources (Resources list.) Wiki page edited by francois.parmentier
  • Dec 04, 2008
    issue 12 (Make License a module to import) reported by francois.parmentier   -   What steps will reproduce the problem? 1. Go to another directory than the root directory of the project 2. Launch pyector 3. type @show w or @show c What is the expected output? What do you see instead? One should get the warranty or the copyright. Instead, there will be an error, since the program looks for a file, according to a relative path. Making a module containing a dictionary with lines of the license will replace the opening of a file, using a relative path.
    What steps will reproduce the problem? 1. Go to another directory than the root directory of the project 2. Launch pyector 3. type @show w or @show c What is the expected output? What do you see instead? One should get the warranty or the copyright. Instead, there will be an error, since the program looks for a file, according to a relative path. Making a module containing a dictionary with lines of the license will replace the opening of a file, using a relative path.
  • Dec 04, 2008
    Resources (Resources list.) Wiki page edited by francois.parmentier
  • Dec 04, 2008
    Resources (Resources list.) Wiki page edited by francois.parmentier
  • Dec 04, 2008
    Resources (Resources list.) Wiki page edited by francois.parmentier
  • Dec 03, 2008
    issue 11 (Prevent infinite loop on generating sentence) reported by francois.parmentier   -   Sometimes, when a node links to another node that links to this first node (or even with a longer queue), the generation might not stop. Implement the same mechanism as in cECTOR, which counts the occurrence of a node in the generated sentence, and divides its probability by the occurrence.
    Sometimes, when a node links to another node that links to this first node (or even with a longer queue), the generation might not stop. Implement the same mechanism as in cECTOR, which counts the occurrence of a node in the generated sentence, and divides its probability by the occurrence.
  • Dec 03, 2008
    issue 10 (Link tokens from the generated sentence to the response of t...) Status changed by francois.parmentier   -   r190 fixes this issue.
    Status: Fixed
    r190 fixes this issue.
    Status: Fixed
  • Dec 03, 2008
    r190 (Fixes issue 10. Links the nodes of the generated sentence...) committed by francois.parmentier   -   Fixes issue 10 . Links the nodes of the generated sentence to the node of the next sentence node (in the next iteration, when the next sentence node is known).
    Fixes issue 10 . Links the nodes of the generated sentence to the node of the next sentence node (in the next iteration, when the next sentence node is known).
  • Dec 03, 2008
    r189 (Display co-occurrence of a link.) committed by francois.parmentier   -   Display co-occurrence of a link.
    Display co-occurrence of a link.
  • Dec 03, 2008
    r188 (Add Link.show()) committed by francois.parmentier   -   Add Link.show()
    Add Link.show()
  • Dec 03, 2008
    r187 (Harmonize parameters' names to class name.) committed by francois.parmentier   -   Harmonize parameters' names to class name.
    Harmonize parameters' names to class name.
  • Dec 03, 2008
    r186 (Prepare change in the returned value of generateSentence()) committed by francois.parmentier   -   Prepare change in the returned value of generateSentence()
    Prepare change in the returned value of generateSentence()
  • Nov 28, 2008
    ExpressionCreation (How to create an expression node.) Wiki page added by francois.parmentier
  • Nov 28, 2008
    issue 10 (Link tokens from the generated sentence to the response of t...) reported by francois.parmentier   -   For the moment (v0.3), there is no link between the generated response of Ector to the next user's sentence. In sentence mode, one links the sentence node to the next user's sentence node. But in generated node, the matching sentence node might not exist. So, it would be good to link the nodes from the generated sentence to the next user's sentence node. It is useful, to activate the nodes of the response to a sentence.
    For the moment (v0.3), there is no link between the generated response of Ector to the next user's sentence. In sentence mode, one links the sentence node to the next user's sentence node. But in generated node, the matching sentence node might not exist. So, it would be good to link the nodes from the generated sentence to the next user's sentence node. It is useful, to activate the nodes of the response to a sentence.
  • Nov 28, 2008
    Resources (Resources list.) Wiki page edited by francois.parmentier
  • Nov 26, 2008
    r183 (Update the manual.) committed by francois.parmentier   -   Update the manual.
    Update the manual.
  • Nov 26, 2008
    r182 (Add the beautifySentence() method. It is called when debu...) committed by francois.parmentier   -   Add the beautifySentence() method. It is called when debug mode is off, at the end of generateSentence().
    Add the beautifySentence() method. It is called when debug mode is off, at the end of generateSentence().
  • Nov 26, 2008
    r181 (Make Ector independent from debug global variable.) committed by francois.parmentier   -   Make Ector independent from debug global variable.
    Make Ector independent from debug global variable.
  • Nov 26, 2008
    r180 (Add @debug command.) committed by francois.parmentier   -   Add @debug command.
    Add @debug command.
  • Nov 26, 2008
    issue 9 (Rename Dodger into Masker) Status changed by francois.parmentier   -   r176 fixed the issue.
    Status: Fixed
    r176 fixed the issue.
    Status: Fixed
  • Nov 26, 2008
    r179 (Fix the tests of @bot@ and @user@) committed by francois.parmentier   -   Fix the tests of @bot@ and @user@
    Fix the tests of @bot@ and @user@
 
Hosted by Google Code