Feedback from lecture :(

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
14 messages Options
Reply | Threaded
Open this post in threaded view
|

Feedback from lecture :(

Stéphane Ducasse
students in the lecture on windows downloaded cogvm I do not from where… and used pharo1.3 and
some experienced crashes.

We will have to brand these vms because the situation is getting difficult.
I do not imagine how we can track these problems.

Stef
Reply | Threaded
Open this post in threaded view
|

Re: Feedback from lecture :(

Guido Stepken
Hi!

My suggestion: Drop COGVM. Expecially in the One click Image. Bad for Pharos reputation at that point of development. LLVM is much better choice.


Very interesting to see Squeak/Pharo running in CHROME. With "native client bridge", so objects, data may be handed over to Javascript engine ... 

Makes more sense for me than a isolated VM, no touch, ...

Have fun, Guido Stepken

2012/2/14 Stéphane Ducasse <[hidden email]>
students in the lecture on windows downloaded cogvm I do not from where… and used pharo1.3 and
some experienced crashes.

We will have to brand these vms because the situation is getting difficult.
I do not imagine how we can track these problems.

Stef

Reply | Threaded
Open this post in threaded view
|

Re: Feedback from lecture :(

laurent laffont
In reply to this post by Stéphane Ducasse
Some crashes on Windows here too with Pharo 1.3 and CogVM from Pharo website. Especially when the .image is on a windows network share.

Laurent

On Tue, Feb 14, 2012 at 7:35 PM, Stéphane Ducasse <[hidden email]> wrote:
students in the lecture on windows downloaded cogvm I do not from where… and used pharo1.3 and
some experienced crashes.

We will have to brand these vms because the situation is getting difficult.
I do not imagine how we can track these problems.

Stef

Reply | Threaded
Open this post in threaded view
|

Re: Feedback from lecture :(

Guido Stepken
Known Windows oplocks problem. Disable caching, re-activate oplocks 

2012/2/14 laurent laffont <[hidden email]>
Some crashes on Windows here too with Pharo 1.3 and CogVM from Pharo website. Especially when the .image is on a windows network share.

Laurent


On Tue, Feb 14, 2012 at 7:35 PM, Stéphane Ducasse <[hidden email]> wrote:
students in the lecture on windows downloaded cogvm I do not from where… and used pharo1.3 and
some experienced crashes.

We will have to brand these vms because the situation is getting difficult.
I do not imagine how we can track these problems.

Stef


Reply | Threaded
Open this post in threaded view
|

Re: Feedback from lecture :(

Mariano Martinez Peck
In reply to this post by Stéphane Ducasse


On Tue, Feb 14, 2012 at 7:35 PM, Stéphane Ducasse <[hidden email]> wrote:
students in the lecture on windows downloaded cogvm I do not from where… and used pharo1.3 and
some experienced crashes.

This has been reported at least 10 times per week since the last 6 months.
The VMs of the one click are FUC.. old and buggy. If nobody have time to make a new one click (we are all busy), please please please, remove it from the website. We give so much bad impression. It is worst than having no one click.
 

We will have to brand these vms because the situation is getting difficult.

No, that's a second step. Put it in Jenkins is the third. Do the first step: do a new OneClick BY HAND, with latest Cog, even without "pharo blessing". That way new comers will not run away from pharo beacuse it crashes.


 
I do not imagine how we can track these problems.

Stef



--
Mariano
http://marianopeck.wordpress.com

Reply | Threaded
Open this post in threaded view
|

Re: Feedback from lecture :(

Igor Stasenko
In reply to this post by Guido Stepken
On 14 February 2012 19:50, Guido Stepken <[hidden email]> wrote:

> Hi!
>
> My suggestion: Drop COGVM. Expecially in the One click Image. Bad for Pharos
> reputation at that point of development. LLVM is much better choice.
>
> http://www.youtube.com/watch?v=W7-Mjx6khr8
>
> Very interesting to see Squeak/Pharo running in CHROME. With "native client
> bridge", so objects, data may be handed over to Javascript engine ...
>
> Makes more sense for me than a isolated VM, no touch, ...
>

Hello again. In another thread. :)

Now just for the record, what you see here?
You see a Squeak VM, compiled for NaCl.
Yes, the very same VM which is used by Pharo.
You can also discover that very same VM is run on various
devices/environments without much changes.
Now you proposing to drop it, and use what instead? LLVM?
What is LLVM?
Here , read:

The LLVM Project is a collection of modular and reusable compiler and
toolchain technologies. Despite its name, LLVM has little to do with
traditional virtual machines, though it does provide helpful libraries
that can be used to build them.
---
So, it means that someone has to reimplement the whole thing on top of
LLVM, just to make the very same thing again, which we having already.
Good idea. Now go and do it.

> Have fun, Guido Stepken
>
>
> 2012/2/14 Stéphane Ducasse <[hidden email]>
>>
>> students in the lecture on windows downloaded cogvm I do not from where…
>> and used pharo1.3 and
>> some experienced crashes.
>>
>> We will have to brand these vms because the situation is getting
>> difficult.
>> I do not imagine how we can track these problems.
>>
>> Stef
>
>



--
Best regards,
Igor Stasenko.

Reply | Threaded
Open this post in threaded view
|

Re: Feedback from lecture :(

Lukas Renggli
In reply to this post by Mariano Martinez Peck
>> students in the lecture on windows downloaded cogvm I do not from where…
>> and used pharo1.3 and
>> some experienced crashes.
>
> This has been reported at least 10 times per week since the last 6 months.
> The VMs of the one click are FUC.. old and buggy. If nobody have time to
> make a new one click (we are all busy), please please please, remove it from
> the website. We give so much bad impression. It is worst than having no one
> click.

You cannot say that, because I updated the One-Click distribution less
than a month ago:

   https://github.com/renggli/builder/commit/7623726096db2e3fa58eefe267c62ba171049a83

People using this code get the latest available CogVM. The problem is
that Pharo maintains its own unmaintained fork.

Lukas

--
Lukas Renggli
www.lukas-renggli.ch

Reply | Threaded
Open this post in threaded view
|

Re: Feedback from lecture :(

Stéphane Ducasse
In reply to this post by Mariano Martinez Peck

> On Tue, Feb 14, 2012 at 7:35 PM, Stéphane Ducasse <[hidden email]> wrote:
> students in the lecture on windows downloaded cogvm I do not from where… and used pharo1.3 and
> some experienced crashes.
>
> This has been reported at least 10 times per week since the last 6 months.
> The VMs of the one click are FUC.. old and buggy. If nobody have time to make a new one click (we are all busy), please please please, remove it from the website. We give so much bad impression. It is worst than having no one click.
>  
>
> We will have to brand these vms because the situation is getting difficult.
>
> No, that's a second step. Put it in Jenkins is the third. Do the first step: do a new OneClick BY HAND, with latest Cog, even without "pharo blessing". That way new comers will not run away from pharo beacuse it crashes.

Where can I find the correct VMs?
I can give a try.
Now I did not do it before and I'm really busy.


>
>
>  
> I do not imagine how we can track these problems.
>
> Stef
>
>
>
> --
> Mariano
> http://marianopeck.wordpress.com
>


Reply | Threaded
Open this post in threaded view
|

Re: Feedback from lecture :(

Igor Stasenko
In reply to this post by Mariano Martinez Peck
On 14 February 2012 20:04, Mariano Martinez Peck <[hidden email]> wrote:

>
>
> On Tue, Feb 14, 2012 at 7:35 PM, Stéphane Ducasse
> <[hidden email]> wrote:
>>
>> students in the lecture on windows downloaded cogvm I do not from where…
>> and used pharo1.3 and
>> some experienced crashes.
>
>
> This has been reported at least 10 times per week since the last 6 months.
> The VMs of the one click are FUC.. old and buggy. If nobody have time to
> make a new one click (we are all busy), please please please, remove it from
> the website. We give so much bad impression. It is worst than having no one
> click.
>

It was known before we released one-click, that VM is not perfect.
But we cannot delay the release because of that. It would make things
only worse.

Because every time you going to release, you can find one or another issue which
needs to be fixed before release.
What is important to understand, that we don't releasing a final
product, the Pharo releases
 are milestones on continuing development.
Yes, every release having issues and they cannot be perfect ,and there
are always will be things
which left not done/not fixed etc etc. So we need to accept that.


>>
>>
>> We will have to brand these vms because the situation is getting
>> difficult.
>
>
> No, that's a second step. Put it in Jenkins is the third. Do the first step:
> do a new OneClick BY HAND, with latest Cog, even without "pharo blessing".
> That way new comers will not run away from pharo beacuse it crashes.
>

And why you think that new one will be more stable that old?
i just recently built a windows VM.
it crashes occasionally/randomly in some bitblt primitive(s) during
drawing a progress bar
when installing MC package, i have no idea why.
I suspect a memory corruption somewhere when new classes installed in
a system , or
existing classes are modified by MC, and bitblt prim is just a
consequence of it.

Which not saying that we should not address this issue(s), but that
latest version is not
necessary more stable than older one.
To guarantee that there are no regressions like that, we need to setup
good testing infrastructure around VM.
And while we're still not there.. but will be soon.


>
>
>>
>> I do not imagine how we can track these problems.
>>
>> Stef
>
>
>
>
> --
> Mariano
> http://marianopeck.wordpress.com
>



--
Best regards,
Igor Stasenko.

Reply | Threaded
Open this post in threaded view
|

Re: Feedback from lecture :(

Stéphane Ducasse
In reply to this post by Lukas Renggli

On Feb 14, 2012, at 8:22 PM, Lukas Renggli wrote:

>>> students in the lecture on windows downloaded cogvm I do not from where…
>>> and used pharo1.3 and
>>> some experienced crashes.
>>
>> This has been reported at least 10 times per week since the last 6 months.
>> The VMs of the one click are FUC.. old and buggy. If nobody have time to
>> make a new one click (we are all busy), please please please, remove it from
>> the website. We give so much bad impression. It is worst than having no one
>> click.
>
> You cannot say that, because I updated the One-Click distribution less
> than a month ago:
>
>   https://github.com/renggli/builder/commit/7623726096db2e3fa58eefe267c62ba171049a83
>
> People using this code get the latest available CogVM. The problem is
> that Pharo maintains its own unmaintained fork.

It is not unmaintained. Now if really few people help then igor cannot do everything.
And since eliot does not care about the effort we spent to make sure that people can compile
their own vms then we have to life with that.
Also the pharo vms has important fixes for us.


> Lukas
>
> --
> Lukas Renggli
> www.lukas-renggli.ch
>


Reply | Threaded
Open this post in threaded view
|

Re: Feedback from lecture :(

Guido Stepken
In reply to this post by Igor Stasenko


Am 14.02.2012 um 20:10 schrieb Igor Stasenko

> What is LLVM?
> Here , read:
>
> The LLVM Project is a collection of modular and reusable compiler and
> toolchain technologies. Despite its name, LLVM has little to do with
> traditional virtual machines, though it does provide helpful libraries
> that can be used to build them.

Read: http://fosdem.org/2012/interview/david-chisnall

tnx 4 understanding, Guido Stepken

Reply | Threaded
Open this post in threaded view
|

Re: Feedback from lecture :(

Igor Stasenko
In reply to this post by Lukas Renggli
On 14 February 2012 20:22, Lukas Renggli <[hidden email]> wrote:

>>> students in the lecture on windows downloaded cogvm I do not from where…
>>> and used pharo1.3 and
>>> some experienced crashes.
>>
>> This has been reported at least 10 times per week since the last 6 months.
>> The VMs of the one click are FUC.. old and buggy. If nobody have time to
>> make a new one click (we are all busy), please please please, remove it from
>> the website. We give so much bad impression. It is worst than having no one
>> click.
>
> You cannot say that, because I updated the One-Click distribution less
> than a month ago:
>
>   https://github.com/renggli/builder/commit/7623726096db2e3fa58eefe267c62ba171049a83
>
> People using this code get the latest available CogVM. The problem is
> that Pharo maintains its own unmaintained fork.
>

Huh?

See the build history:
https://ci.lille.inria.fr/pharo/view/Cog/job/Cog-Unix/

last build 26 Jan

https://ci.lille.inria.fr/pharo/view/Cog/job/StackVM-Mac-Cocoa-blessed/
again, last build 28 Jan.

Here just a list of changes in latest builds:
(you can find them here:
https://ci.lille.inria.fr/pharo/view/Cog/job/Cog%20Git%20Tracker%20%28blessed%29/changes
)
Changes
#26 (Jan 26, 2012 5:33:21 PM)

    replaced spaces with tabs — camillobruni / detail

#25 (Jan 26, 2012 5:18:48 PM)

    added new pharo scropt — camillobruni / detail
    - fixed configuraiton to open a Workspace with example
configurations — camillobruni / detail
    trying to fix issues with the command line arguments — camillobruni / detail
    more work on the image script and the howto — camillobruni / detail
    tried to implemention dir_lookupEntry for Cococa OSX — camillobruni / detail
    extended howto with osx gcc change — camillobruni / detail
    fixed back opengl changes — camillobruni / detail
    fixed VM path, returning bundlepath with app name — camillobruni / detail
    implemented direEntryLookup and refactored a bit the source code —
camillobruni / detail
    more howto — camillobruni / detail
    minor change in the HOWTO — camillobruni / detail
    undoing accidental changes — camillobruni / detail

#23 (Jan 25, 2012 8:06:27 PM)

    removed deprecated calls to recordKeystroke — guillermopolito / detail
    removed unused comment — guillermopolito / detail
    removed old code — guillermopolito / detail
    removed dead code — guillermopolito / detail
    removed dead code — guillermopolito / detail
    removed lot of debug only code — guillermopolito / detail
    removed temporal file — guillermopolito / detail
    removed more debug printfs — guillermopolito / detail
    formatted the code — guillermopolito / detail
    little style change — guillermopolito / detail
    added mappings to numeric keypad — guillermopolito / detail
    added comments to unicode conversion — guillermopolito / detail
    more comments — guillermopolito / detail

For the full list of changes in VM , visit
https://gitorious.org/cogvm/blessed/commits/master

but just don't say bullshit like unmaintained unmaintained VM.

Now if people do not downloading it, do not report problems, and
sticking to Eliot's VMs, it makes no sense
to do anything.
You know well, that we cannot force people to use it or not.

(i triggered new build now just to check if everything ok).

--
Best regards,
Igor Stasenko.

Reply | Threaded
Open this post in threaded view
|

Re: Feedback from lecture :(

Eliot Miranda-2


On Tue, Feb 14, 2012 at 11:52 AM, Igor Stasenko <[hidden email]> wrote:
On 14 February 2012 20:22, Lukas Renggli <[hidden email]> wrote:
>>> students in the lecture on windows downloaded cogvm I do not from where…
>>> and used pharo1.3 and
>>> some experienced crashes.
>>
>> This has been reported at least 10 times per week since the last 6 months.
>> The VMs of the one click are FUC.. old and buggy. If nobody have time to
>> make a new one click (we are all busy), please please please, remove it from
>> the website. We give so much bad impression. It is worst than having no one
>> click.
>
> You cannot say that, because I updated the One-Click distribution less
> than a month ago:
>
>   https://github.com/renggli/builder/commit/7623726096db2e3fa58eefe267c62ba171049a83
>
> People using this code get the latest available CogVM. The problem is
> that Pharo maintains its own unmaintained fork.
>

Huh?

See the build history:
https://ci.lille.inria.fr/pharo/view/Cog/job/Cog-Unix/

last build 26 Jan

https://ci.lille.inria.fr/pharo/view/Cog/job/StackVM-Mac-Cocoa-blessed/
again, last build 28 Jan.

Here just a list of changes in latest builds:
(you can find them here:
https://ci.lille.inria.fr/pharo/view/Cog/job/Cog%20Git%20Tracker%20%28blessed%29/changes
)
Changes
#26 (Jan 26, 2012 5:33:21 PM)

   replaced spaces with tabs — camillobruni / detail

#25 (Jan 26, 2012 5:18:48 PM)

   added new pharo scropt — camillobruni / detail
   - fixed configuraiton to open a Workspace with example
configurations — camillobruni / detail
   trying to fix issues with the command line arguments — camillobruni / detail
   more work on the image script and the howto — camillobruni / detail
   tried to implemention dir_lookupEntry for Cococa OSX — camillobruni / detail
   extended howto with osx gcc change — camillobruni / detail
   fixed back opengl changes — camillobruni / detail
   fixed VM path, returning bundlepath with app name — camillobruni / detail
   implemented direEntryLookup and refactored a bit the source code —
camillobruni / detail
   more howto — camillobruni / detail
   minor change in the HOWTO — camillobruni / detail
   undoing accidental changes — camillobruni / detail

#23 (Jan 25, 2012 8:06:27 PM)

   removed deprecated calls to recordKeystroke — guillermopolito / detail
   removed unused comment — guillermopolito / detail
   removed old code — guillermopolito / detail
   removed dead code — guillermopolito / detail
   removed dead code — guillermopolito / detail
   removed lot of debug only code — guillermopolito / detail
   removed temporal file — guillermopolito / detail
   removed more debug printfs — guillermopolito / detail
   formatted the code — guillermopolito / detail
   little style change — guillermopolito / detail
   added mappings to numeric keypad — guillermopolito / detail
   added comments to unicode conversion — guillermopolito / detail
   more comments — guillermopolito / detail

For the full list of changes in VM , visit
https://gitorious.org/cogvm/blessed/commits/master

but just don't say bullshit like unmaintained unmaintained VM.

The substantive change is the ephemeron changes.  I'm not convinced that this code is correct.  I'm convinced that the approach is wrong.  See my postings on using the format word to encode ephemeronness.
 

Now if people do not downloading it, do not report problems, and
sticking to Eliot's VMs, it makes no sense
to do anything.
You know well, that we cannot force people to use it or not.

(i triggered new build now just to check if everything ok).

--
Best regards,
Igor Stasenko.




--
best,
Eliot

Reply | Threaded
Open this post in threaded view
|

Re: Feedback from lecture :(

Igor Stasenko
On 15 February 2012 00:03, Eliot Miranda <[hidden email]> wrote:

>
>
> On Tue, Feb 14, 2012 at 11:52 AM, Igor Stasenko <[hidden email]> wrote:
>>
>> On 14 February 2012 20:22, Lukas Renggli <[hidden email]> wrote:
>> >>> students in the lecture on windows downloaded cogvm I do not from
>> >>> where…
>> >>> and used pharo1.3 and
>> >>> some experienced crashes.
>> >>
>> >> This has been reported at least 10 times per week since the last 6
>> >> months.
>> >> The VMs of the one click are FUC.. old and buggy. If nobody have time
>> >> to
>> >> make a new one click (we are all busy), please please please, remove it
>> >> from
>> >> the website. We give so much bad impression. It is worst than having no
>> >> one
>> >> click.
>> >
>> > You cannot say that, because I updated the One-Click distribution less
>> > than a month ago:
>> >
>> >
>> > https://github.com/renggli/builder/commit/7623726096db2e3fa58eefe267c62ba171049a83
>> >
>> > People using this code get the latest available CogVM. The problem is
>> > that Pharo maintains its own unmaintained fork.
>> >
>>
>> Huh?
>>
>> See the build history:
>> https://ci.lille.inria.fr/pharo/view/Cog/job/Cog-Unix/
>>
>> last build 26 Jan
>>
>> https://ci.lille.inria.fr/pharo/view/Cog/job/StackVM-Mac-Cocoa-blessed/
>> again, last build 28 Jan.
>>
>> Here just a list of changes in latest builds:
>> (you can find them here:
>>
>> https://ci.lille.inria.fr/pharo/view/Cog/job/Cog%20Git%20Tracker%20%28blessed%29/changes
>> )
>> Changes
>> #26 (Jan 26, 2012 5:33:21 PM)
>>
>>    replaced spaces with tabs — camillobruni / detail
>>
>> #25 (Jan 26, 2012 5:18:48 PM)
>>
>>    added new pharo scropt — camillobruni / detail
>>    - fixed configuraiton to open a Workspace with example
>> configurations — camillobruni / detail
>>    trying to fix issues with the command line arguments — camillobruni /
>> detail
>>    more work on the image script and the howto — camillobruni / detail
>>    tried to implemention dir_lookupEntry for Cococa OSX — camillobruni /
>> detail
>>    extended howto with osx gcc change — camillobruni / detail
>>    fixed back opengl changes — camillobruni / detail
>>    fixed VM path, returning bundlepath with app name — camillobruni /
>> detail
>>    implemented direEntryLookup and refactored a bit the source code —
>> camillobruni / detail
>>    more howto — camillobruni / detail
>>    minor change in the HOWTO — camillobruni / detail
>>    undoing accidental changes — camillobruni / detail
>>
>> #23 (Jan 25, 2012 8:06:27 PM)
>>
>>    removed deprecated calls to recordKeystroke — guillermopolito / detail
>>    removed unused comment — guillermopolito / detail
>>    removed old code — guillermopolito / detail
>>    removed dead code — guillermopolito / detail
>>    removed dead code — guillermopolito / detail
>>    removed lot of debug only code — guillermopolito / detail
>>    removed temporal file — guillermopolito / detail
>>    removed more debug printfs — guillermopolito / detail
>>    formatted the code — guillermopolito / detail
>>    little style change — guillermopolito / detail
>>    added mappings to numeric keypad — guillermopolito / detail
>>    added comments to unicode conversion — guillermopolito / detail
>>    more comments — guillermopolito / detail
>>
>> For the full list of changes in VM , visit
>> https://gitorious.org/cogvm/blessed/commits/master
>>
>> but just don't say bullshit like unmaintained unmaintained VM.
>
>
> The substantive change is the ephemeron changes.  I'm not convinced that
> this code is correct.  I'm convinced that the approach is wrong.  See my
> postings on using the format word to encode ephemeronness.
>

I remember that, Eliot. And i agreed with you that it is not optimal.
You also told back then, that you will introduce it soon,
so i suspended any activities in that regard.

I can revert that code from VM in no time, if you think it is the one,
which makes VM unstable.
Which i doubt, since ephemerons is not in use by image, so the code in
question simply does an empty run, resulting same
semantics as in your branch.
But i will check it once more.

--
Best regards,
Igor Stasenko.