[Pharo-project] Time Profile mostly unusable
In Pharo 1.3, the new Time Profile is unusable (the one you get from profiling the UI): - In the resulting window clicking in a line cause an error - WORST: there is no way (at least obvious one) to copy and past the resulting profile in an external file. When replacing the tools, beware to replace with working ones otherwise developers may run away. I am updating my image to see if it is a temporary problem. Hilaire -- Education 0.2 -- http://blog.ofset.org/hilaire
Hi Hilaire please, open an issue. Cheers Alain Le 21/07/2011 13:48, Hilaire Fernandes a écrit :
In Pharo 1.3, the new Time Profile is unusable (the one you get from profiling the UI):
- In the resulting window clicking in a line cause an error - WORST: there is no way (at least obvious one) to copy and past the resulting profile in an external file.
When replacing the tools, beware to replace with working ones otherwise developers may run away.
I am updating my image to see if it is a temporary problem.
Hilaire
it depends on your scenario and when users do not use or use the tool differently you get different feedback. Stef On Jul 21, 2011, at 1:48 PM, Hilaire Fernandes wrote:
In Pharo 1.3, the new Time Profile is unusable (the one you get from profiling the UI):
- In the resulting window clicking in a line cause an error - WORST: there is no way (at least obvious one) to copy and past the resulting profile in an external file.
When replacing the tools, beware to replace with working ones otherwise developers may run away.
I am updating my image to see if it is a temporary problem.
Hilaire
-- Education 0.2 -- http://blog.ofset.org/hilaire
On Thu, Jul 21, 2011 at 6:14 PM, Stéphane Ducasse <stephane.ducasse@inria.fr
wrote:
it depends on your scenario and when users do not use or use the tool differently you get different feedback.
Or even better: if noboy says anything in the last 6 months while Pharo 1.3 was being developed/tested then we cannot do magic. I am tired of saying that anything that YOU don't test, can be broken.
Stef
On Jul 21, 2011, at 1:48 PM, Hilaire Fernandes wrote:
In Pharo 1.3, the new Time Profile is unusable (the one you get from profiling the UI):
- In the resulting window clicking in a line cause an error - WORST: there is no way (at least obvious one) to copy and past the resulting profile in an external file.
When replacing the tools, beware to replace with working ones otherwise developers may run away.
I am updating my image to see if it is a temporary problem.
Hilaire
-- Education 0.2 -- http://blog.ofset.org/hilaire
-- Mariano http://marianopeck.wordpress.com
On 21 July 2011 20:29, Mariano Martinez Peck <marianopeck@gmail.com> wrote:
On Thu, Jul 21, 2011 at 6:14 PM, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:
it depends on your scenario and when users do not use or use the tool differently you get different feedback.
Or even better: if noboy says anything in the last 6 months while Pharo 1.3 was being developed/tested then we cannot do magic. I am tired of saying that anything that YOU don't test, can be broken.
why "can be"? it definitely will be broken. What would be nice is to write tests for TimeProfileBrowser to prevent regressions. Any ideas? -- Best regards, Igor Stasenko AKA sig.
this is mainly a ui concern. On Jul 21, 2011, at 8:58 PM, Igor Stasenko wrote:
On 21 July 2011 20:29, Mariano Martinez Peck <marianopeck@gmail.com> wrote:
On Thu, Jul 21, 2011 at 6:14 PM, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:
it depends on your scenario and when users do not use or use the tool differently you get different feedback.
Or even better: if noboy says anything in the last 6 months while Pharo 1.3 was being developed/tested then we cannot do magic. I am tired of saying that anything that YOU don't test, can be broken.
why "can be"? it definitely will be broken.
What would be nice is to write tests for TimeProfileBrowser to prevent regressions. Any ideas?
-- Best regards, Igor Stasenko AKA sig.
I was able to write some pretty reasonable tests for gui elements in Dolphin. Getting something to open/close and in between, firing a few commands at it can be informative, sometimes ~:0 ________________________________________ From: pharo-project-bounces@lists.gforge.inria.fr [pharo-project-bounces@lists.gforge.inria.fr] On Behalf Of Stéphane Ducasse [stephane.ducasse@inria.fr] Sent: Thursday, July 21, 2011 3:09 PM To: Pharo-project@lists.gforge.inria.fr Subject: Re: [Pharo-project] Time Profile mostly unusable this is mainly a ui concern. On Jul 21, 2011, at 8:58 PM, Igor Stasenko wrote:
On 21 July 2011 20:29, Mariano Martinez Peck <marianopeck@gmail.com> wrote:
On Thu, Jul 21, 2011 at 6:14 PM, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:
it depends on your scenario and when users do not use or use the tool differently you get different feedback.
Or even better: if noboy says anything in the last 6 months while Pharo 1.3 was being developed/tested then we cannot do magic. I am tired of saying that anything that YOU don't test, can be broken.
why "can be"? it definitely will be broken.
What would be nice is to write tests for TimeProfileBrowser to prevent regressions. Any ideas?
-- Best regards, Igor Stasenko AKA sig.
Given the amount of changes in Pharo it is normal stuffs get broken and some tests need to be human conducted, therefore a list of features to test before release could be used to review the needed working features. Wiki shared, the workload on this list can be easily contributed by anyone. I just see problem in the the version tool of the browser: - the tool does not show the diff properly: no way to read what is new and old - the revert just does not work May be the rate of release could be slow down to get properly working Image, no ? Regarding the problem with the Time Profile, I see a patch from Bert, but upgrading my image just broke, so I can't test it for now. Hilaire Le 21/07/2011 21:09, Stéphane Ducasse a écrit :
this is mainly a ui concern.
On Jul 21, 2011, at 8:58 PM, Igor Stasenko wrote:
On 21 July 2011 20:29, Mariano Martinez Peck <marianopeck@gmail.com> wrote:
On Thu, Jul 21, 2011 at 6:14 PM, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:
it depends on your scenario and when users do not use or use the tool differently you get different feedback.
Or even better: if noboy says anything in the last 6 months while Pharo 1.3 was being developed/tested then we cannot do magic. I am tired of saying that anything that YOU don't test, can be broken.
why "can be"? it definitely will be broken.
What would be nice is to write tests for TimeProfileBrowser to prevent regressions. Any ideas?
-- Best regards, Igor Stasenko AKA sig.
-- Education 0.2 -- http://blog.ofset.org/hilaire
Given the amount of changes in Pharo it is normal stuffs get broken and some tests need to be human conducted, therefore a list of features to test before release could be used to review the needed working features. Wiki shared, the workload on this list can be easily contributed by anyone.
I just see problem in the the version tool of the browser: - the tool does not show the diff properly: no way to read what is new and old
where?
- the revert just does not work
do you have a scenario
May be the rate of release could be slow down to get properly working Image, no ?
Not really. When you pay a person for 3 months of work you cannot ask him to work for a year slowler. We payed benjamin for a couple of months and after we can just be happy that he is doing post actions especially when the semester will restart it will be more difficult.
Regarding the problem with the Time Profile, I see a patch from Bert, but upgrading my image just broke, so I can't test it for now.
Which patch? Where is the bug entry? On which image? breaking where? Stef
Hilaire
Le 21/07/2011 21:09, Stéphane Ducasse a écrit :
this is mainly a ui concern.
On Jul 21, 2011, at 8:58 PM, Igor Stasenko wrote:
On 21 July 2011 20:29, Mariano Martinez Peck <marianopeck@gmail.com> wrote:
On Thu, Jul 21, 2011 at 6:14 PM, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote:
it depends on your scenario and when users do not use or use the tool differently you get different feedback.
Or even better: if noboy says anything in the last 6 months while Pharo 1.3 was being developed/tested then we cannot do magic. I am tired of saying that anything that YOU don't test, can be broken.
why "can be"? it definitely will be broken.
What would be nice is to write tests for TimeProfileBrowser to prevent regressions. Any ideas?
-- Best regards, Igor Stasenko AKA sig.
-- Education 0.2 -- http://blog.ofset.org/hilaire
Le 22/07/2011 10:58, Stéphane Ducasse a écrit :
Given the amount of changes in Pharo it is normal stuffs get broken and some tests need to be human conducted, therefore a list of features to test before release could be used to review the needed working features. Wiki shared, the workload on this list can be easily contributed by anyone.
I just see problem in the the version tool of the browser: - the tool does not show the diff properly: no way to read what is new and old
where?
in the version tool of the class browser
- the revert just does not work
do you have a scenario
Clic the button 'version' in the class browser.
May be the rate of release could be slow down to get properly working Image, no ?
Not really. When you pay a person for 3 months of work you cannot ask him to work for a year slowler. We payed benjamin for a couple of months and after we can just be happy that he is doing post actions especially when the semester will restart it will be more difficult.
My remarks were about release not the written code or added feature.
Regarding the problem with the Time Profile, I see a patch from Bert, but upgrading my image just broke, so I can't test it for now.
Which patch? Where is the bug entry?
In bugs tracker, search for closed issue with keyword Time Profile
On which image? breaking where?
See my first email in this thread. I double check, the inheritence is broken from the Browser. Scenario: clic the inhéritence button. I will check with latest 1.3 with up to date update before adding entry in the bugs tracker. Now I have to finish some other works first. Hilaire
Le 22/07/2011 15:47, Hilaire Fernandes a écrit :
I will check with latest 1.3 with up to date update before adding entry in the bugs tracker. Now I have to finish some other works first.
Here are my tests: - version works ok in last 1.3 - time profiler fixed, but profile2Text needed: http://code.google.com/p/pharo/issues/detail?id=4546 - inheritance in the browser broken: http://code.google.com/p/pharo/issues/detail?id=4545 Hilaire -- Education 0.2 -- http://blog.ofset.org/hilaire
On Jul 22, 2011, at 4:35 PM, Hilaire Fernandes wrote:
Le 22/07/2011 15:47, Hilaire Fernandes a écrit :
I will check with latest 1.3 with up to date update before adding entry in the bugs tracker. Now I have to finish some other works first.
Here are my tests:
- version works ok in last 1.3
- time profiler fixed, but profile2Text needed: http://code.google.com/p/pharo/issues/detail?id=4546
- inheritance in the browser broken: http://code.google.com/p/pharo/issues/detail?id=4545
sorry but I guess that in 1.3 you will not have that. May be we should ask for implementors instead. We will rewrite all the tools and the results will be different. We want to avoid to have to maintain two or three times the same behavior especially with the quality of the old code. Stef
Le 22/07/2011 22:17, Stéphane Ducasse a écrit :
- inheritance in the browser broken: http://code.google.com/p/pharo/issues/detail?id=4545
sorry but I guess that in 1.3 you will not have that. May be we should ask for implementors instead. We will rewrite all the tools and the results will be different. We want to avoid to have to maintain two or three times the same behavior especially with the quality of the old code.
That's ok but beware enduser (developer) will not specially want to know about the detail about why it does not work. May be 1.3 should be marked as an intermediate release, not for production or things like that. Hilaire -- Education 0.2 -- http://blog.ofset.org/hilaire
On Jul 23, 2011, at 9:40 AM, Hilaire Fernandes wrote:
Le 22/07/2011 22:17, Stéphane Ducasse a écrit :
- inheritance in the browser broken: http://code.google.com/p/pharo/issues/detail?id=4545
sorry but I guess that in 1.3 you will not have that. May be we should ask for implementors instead. We will rewrite all the tools and the results will be different. We want to avoid to have to maintain two or three times the same behavior especially with the quality of the old code.
That's ok but beware enduser (developer) will not specially want to know about the detail about why it does not work.
sorry but open-source dev is not like off the shelf tools. Knowing why things change are also important. We are building in addition to software a social community.
May be 1.3 should be marked as an intermediate release, not for production or things like that.
it is not an intermediate release
Hilaire
-- Education 0.2 -- http://blog.ofset.org/hilaire
Le 23/07/2011 11:01, Stéphane Ducasse a écrit :
That's ok but beware enduser (developer) will not specially want to know about the detail about why it does not work.
sorry but open-source dev is not like off the shelf tools. Knowing why things change are also important. We are building in addition to software a social community.
Sure, but the users in opensource are also very volatile, I know by experience how hard it can beat you. Hilaire -- Education 0.2 -- http://blog.ofset.org/hilaire
Am 21.07.2011 um 20:29 schrieb Mariano Martinez Peck:
On Thu, Jul 21, 2011 at 6:14 PM, Stéphane Ducasse <stephane.ducasse@inria.fr> wrote: it depends on your scenario and when users do not use or use the tool differently you get different feedback.
Or even better: if noboy says anything in the last 6 months while Pharo 1.3 was being developed/tested then we cannot do magic. I am tired of saying that anything that YOU don't test, can be broken.
Seems I learned something today. I never understood TDD as "I develop, YOU test" type of development. Thanks for clarifying. Norbert
Stef
On Jul 21, 2011, at 1:48 PM, Hilaire Fernandes wrote:
In Pharo 1.3, the new Time Profile is unusable (the one you get from profiling the UI):
- In the resulting window clicking in a line cause an error - WORST: there is no way (at least obvious one) to copy and past the resulting profile in an external file.
When replacing the tools, beware to replace with working ones otherwise developers may run away.
I am updating my image to see if it is a temporary problem.
Hilaire
-- Education 0.2 -- http://blog.ofset.org/hilaire
-- Mariano http://marianopeck.wordpress.com
Le 22/07/2011 10:02, Norbert Hartl a écrit :
Am 21.07.2011 um 20:29 schrieb Mariano Martinez Peck:
On Thu, Jul 21, 2011 at 6:14 PM, Stéphane Ducasse <stephane.ducasse@inria.fr <mailto:stephane.ducasse@inria.fr>> wrote:
it depends on your scenario and when users do not use or use the tool differently you get different feedback.
Or even better: if noboy says anything in the last 6 months while Pharo 1.3 was being developed/tested then we cannot do magic. I am tired of saying that anything that YOU don't test, can be broken.
Seems I learned something today. I never understood TDD as "I develop, YOU test" type of development. Thanks for clarifying.
:) I would say something like "the pharo community develop AND test" I don't know who has implemented this Time Profiler. Now we can USE it for real, report and someone else can fix it and improve it. This is how it works no ? How can we ask newbies to contribute if this is not the case ? How can we make the pharo community grow ? Cheers Alain
Norbert
Stef
On Jul 21, 2011, at 1:48 PM, Hilaire Fernandes wrote:
> In Pharo 1.3, the new Time Profile is unusable (the one you get from > profiling the UI): > > - In the resulting window clicking in a line cause an error > - WORST: there is no way (at least obvious one) to copy and past the > resulting profile in an external file. > > When replacing the tools, beware to replace with working ones otherwise > developers may run away. > > I am updating my image to see if it is a temporary problem. > > Hilaire > > -- > Education 0.2 -- http://blog.ofset.org/hilaire > >
-- Mariano http://marianopeck.wordpress.com <http://marianopeck.wordpress.com/>
Am 22.07.2011 um 15:30 schrieb Alain Plantec:
Le 22/07/2011 10:02, Norbert Hartl a écrit :
Am 21.07.2011 um 20:29 schrieb Mariano Martinez Peck:
On Thu, Jul 21, 2011 at 6:14 PM, Stéphane Ducasse <stephane.ducasse@inria.fr <mailto:stephane.ducasse@inria.fr>> wrote:
it depends on your scenario and when users do not use or use the tool differently you get different feedback.
Or even better: if noboy says anything in the last 6 months while Pharo 1.3 was being developed/tested then we cannot do magic. I am tired of saying that anything that YOU don't test, can be broken.
Seems I learned something today. I never understood TDD as "I develop, YOU test" type of development. Thanks for clarifying.
:) I would say something like "the pharo community develop AND test" I don't know who has implemented this Time Profiler. Now we can USE it for real, report and someone else can fix it and improve it. This is how it works no ? How can we ask newbies to contribute if this is not the case ? How can we make the pharo community grow ?
I don't know. But I think Hillaire had a point in here. There is a change rate a community can cover. As we like to change code it is ok that things break. If complaints about broken things raise than you should see that you overstretched your capability and need to think again. The type of replies I replied to just prove that someone is tired listening. And at least I'm pretty sure that listening is essential. Norbert
Norbert
Stef
On Jul 21, 2011, at 1:48 PM, Hilaire Fernandes wrote:
In Pharo 1.3, the new Time Profile is unusable (the one you get from profiling the UI):
- In the resulting window clicking in a line cause an error - WORST: there is no way (at least obvious one) to copy and past the resulting profile in an external file.
When replacing the tools, beware to replace with working ones otherwise developers may run away.
I am updating my image to see if it is a temporary problem.
Hilaire
-- Education 0.2 -- http://blog.ofset.org/hilaire
-- Mariano http://marianopeck.wordpress.com <http://marianopeck.wordpress.com/>
I don't know. But I think Hillaire had a point in here.
what point? We want to change and rewrite all the tools. Now we can only do that with people working. Now we remove the inheritance class browser. and now if you want to see the implementor of add: then you do not use inheritance anymore. You should use implementors or you help in nautilus and rpackage because nautilus implement inheritance.
There is a change rate a community can cover. As we like to change code it is ok that things break. If complaints about broken things raise than you should see that you overstretched your capability and need to think again.
Yes but I do not think that we are in this situation. The point is to consider the ratio of changes vs bugs. I can pay Igor one and half year after igor may be forced to sell his PHP skills again :(. So we can go slowly and play cards and watch videos, but this means that less will be done. Or we break some stuff from time to time and this is ok and we move at the speed we believe is good. Right now I'm not even working a day a week on Pharo but if I would work more I would not catch more bugs because these "bugs" are stuff that I never use. So if people use different scenario than us, how can we know it without them raising problems? Stef
The type of replies I replied to just prove that someone is tired listening. And at least I'm pretty sure that listening is essential.
Norbert
Am 22.07.2011 um 22:15 schrieb Stéphane Ducasse:
I don't know. But I think Hillaire had a point in here.
what point? We want to change and rewrite all the tools. Now we can only do that with people working. Now we remove the inheritance class browser. and now if you want to see the implementor of add: then you do not use inheritance anymore. You should use implementors or you help in nautilus and rpackage because nautilus implement inheritance.
There is a change rate a community can cover. As we like to change code it is ok that things break. If complaints about broken things raise than you should see that you overstretched your capability and need to think again.
Yes but I do not think that we are in this situation.
Ok.
The point is to consider the ratio of changes vs bugs.
Well, that's what I'm saying.
I can pay Igor one and half year after igor may be forced to sell his PHP skills again :(. So we can go slowly and play cards and watch videos, but this means that less will be done.
Or we break some stuff from time to time and this is ok and we move at the speed we believe is good. Right now I'm not even working a day a week on Pharo but if I would work more I would not catch more bugs because these "bugs" are stuff that I never use. So if people use different scenario than us, how can we know it without them raising problems?
Thanks Stef for always answering verbose :) I can see that you have more typical kind of project management problems combined with a community. I don't want to talk against anything. In the last weeks I just heard a lot of complaints from some "more core" developers about complaints from others "not so core". And I can't see much sense in "complaining about complaining". That's all. Norbert
Stef
The type of replies I replied to just prove that someone is tired listening. And at least I'm pretty sure that listening is essential.
Norbert
A final word on this thread. We are not changing for the sake of it. With time we are just getting to lower parts of the system. I imagine that slowly we will arrive to a fixpoint in certain areas. And this is good. I would like to be able to focus on my research and build cool stuff on top of pharo. Now we are working on a roadmap slideshow to show where we are working and what is on our radar so that you can understand it. Stef
Am 23.07.2011 um 20:51 schrieb Stéphane Ducasse <stephane.ducasse@inria.fr>:
A final word on this thread. We are not changing for the sake of it. With time we are just getting to lower parts of the system. I imagine that slowly we will arrive to a fixpoint in certain areas. And this is good. I would like to be able to focus on my research and build cool stuff on top of pharo. Now we are working on a roadmap slideshow to show where we are working and what is on our radar so that you can understand it.
great! Norbert
Don't worry, we are already building cool stuff with Pharo! Hilaire Le 23/07/2011 20:51, Stéphane Ducasse a écrit :
A final word on this thread. We are not changing for the sake of it. With time we are just getting to lower parts of the system. I imagine that slowly we will arrive to a fixpoint in certain areas. And this is good. I would like to be able to focus on my research and build cool stuff on top of pharo. Now we are working on a roadmap slideshow to show where we are working and what is on our radar so that you can understand it.
Stef
-- Education 0.2 -- http://blog.ofset.org/hilaire
participants (7)
-
Alain Plantec -
Hilaire Fernandes -
Igor Stasenko -
Mariano Martinez Peck -
Norbert Hartl -
Schwab,Wilhelm K -
Stéphane Ducasse