Smalltalk › Frameworks & Tools › SciSmalltalk

first impressions

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

21 messages Options

werner kassens-2

first impressions

Hi,
fwiw i had a look at PMVector:
- asDHBVector : in collection this is renamed to #asPMVector. but this does not exist in PMVector, hence it is not polymorphic any more. #asDHBVector should perhaps be renamed?
- sin : superfluous since PMVector is an Array that understands #sin. further it changes the usual meaning of #sin insofar as it changes the original object instead of returning a new object ( see the difference between #negated and #negate). this is _not_ how pharo deals with functions like that and opens up the possibility of a heck of new bugs.
- cosh, log, sin, sinh, sqrt, tan, tanh: same problem as with sin.
- cumsum : should also return a new object.btw this behaviour is in all (?) cases automatically transferred to matrices which obviously can also produce bugs there.
c:=PMSymmetricMatrix rows:#((1 2)(2 1)).
a:=c cumsum.
c isSymmetric . "true"
c. "a PMVector(1 3)
a PMVector(2 3)"
just a simple example why this general approach in all these functions is problematic.
- sum: is also implemented like sin and superfluous, but here the problem is even worse because the implementation is changed, and not really for the better.
- dot: : #dot: is used in floatarray and means - as expected - "dot product" there. it means something completely else here. you might want to look for another name.
- == : should _not_ be overwritten, see the comment in the overwritten method. additionally it _completely_ changes the meaning of ==, definitely making it unusable.
- <, > : one normally expects a Boolean as a return, not an array, and it is often not mappable in good math systems if i remember correctly. i wonder whether that is really necessary.

well ok, that was PMVector, i better stop now <friendly grin>.
werner

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

Uko

Re: first impressions

I think we should deprecate this one with an automatic rewrite.

Uko

On 17 Mar 2017, at 19:06, werner kassens <[hidden email]> wrote:

#asDHBVector should perhaps be renamed?

werner kassens-2

Re: first impressions

Hi Uko,
i guess a simple renaming would be sufficient, since #asDhbVector has already been replaced by #asPMVector in most cases, there are only two places where this has been forgotten.
werner

On Fri, Mar 17, 2017 at 7:09 PM, Yuriy Tymchuk <[hidden email]> wrote:

I think we should deprecate this one with an automatic rewrite.

Uko

On 17 Mar 2017, at 19:06, werner kassens <[hidden email]> wrote:

#asDHBVector should perhaps be renamed?

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

Uko

Re: first impressions

Yes, I understand, but I’m thinking about the projects that depend on Polymath and I suppose #asDhbVector is a popular method there.

Cheers

Uko

On 18 Mar 2017, at 10:22, werner kassens <[hidden email]> wrote:

Hi Uko,
i guess a simple renaming would be sufficient, since #asDhbVector has already been replaced by #asPMVector in most cases, there are only two places where this has been forgotten.
werner

On Fri, Mar 17, 2017 at 7:09 PM, Yuriy Tymchuk <[hidden email]> wrote:
I think we should deprecate this one with an automatic rewrite.

Uko

On 17 Mar 2017, at 19:06, werner kassens <[hidden email]> wrote:

#asDHBVector should perhaps be renamed?

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

werner kassens-2

Re: first impressions

yes Uko, i agree it is a popular method, you are probably right.

werner

On Sat, Mar 18, 2017 at 10:32 AM, Yuriy Tymchuk <[hidden email]> wrote:

Yes, I understand, but I’m thinking about the projects that depend on Polymath and I suppose #asDhbVector is a popular method there.

Cheers
Uko

On 18 Mar 2017, at 10:22, werner kassens <[hidden email]> wrote:

Hi Uko,
i guess a simple renaming would be sufficient, since #asDhbVector has already been replaced by #asPMVector in most cases, there are only two places where this has been forgotten.
werner

On Fri, Mar 17, 2017 at 7:09 PM, Yuriy Tymchuk <[hidden email]> wrote:
I think we should deprecate this one with an automatic rewrite.

Uko

On 17 Mar 2017, at 19:06, werner kassens <[hidden email]> wrote:

#asDHBVector should perhaps be renamed?

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

werner kassens-2

Re: first impressions

In reply to this post by werner kassens-2

Hi,
perhaps i should make a little example, why this approach with the mathematical functions is problematic:
a:=2.
a ln + a sin.
"1.6024446073856269"
a1:=a asArray .
"#(2)"
a1 ln + a1 sin.
"#(1.6024446073856269)"
a2:=a1 as: PMVector .
"a PMVector(2)"
a2 ln + a2 sin.
"a PMVector(0.4728027681014221)"

and btw i just had a look at the class side. re ones: & zeros:
in case one finds it too complicated to write
PMVector new: 300 withAll: 0
and prefers to write
PMVector zeros: 300
(something i at least halfway understand), it could perhaps make sense to replace the implementation simply with
^PMVector new: anInteger withAll: 0
it is not only shorter but also faster for bigger numbers (see comment in SequenceableCollection>>atAllPut: )
werner

On Fri, Mar 17, 2017 at 7:06 PM, werner kassens <[hidden email]> wrote:

Hi,
fwiw i had a look at PMVector:
- asDHBVector : in collection this is renamed to #asPMVector. but this does not exist in PMVector, hence it is not polymorphic any more. #asDHBVector should perhaps be renamed?
- sin : superfluous since PMVector is an Array that understands #sin. further it changes the usual meaning of #sin insofar as it changes the original object instead of returning a new object ( see the difference between #negated and #negate). this is _not_ how pharo deals with functions like that and opens up the possibility of a heck of new bugs.
- cosh, log, sin, sinh, sqrt, tan, tanh: same problem as with sin.
- cumsum : should also return a new object.btw this behaviour is in all (?) cases automatically transferred to matrices which obviously can also produce bugs there.
c:=PMSymmetricMatrix rows:#((1 2)(2 1)).
a:=c cumsum.
c isSymmetric . "true"
c. "a PMVector(1 3)
a PMVector(2 3)"
just a simple example why this general approach in all these functions is problematic.
- sum: is also implemented like sin and superfluous, but here the problem is even worse because the implementation is changed, and not really for the better.
- dot: : #dot: is used in floatarray and means - as expected - "dot product" there. it means something completely else here. you might want to look for another name.
- == : should _not_ be overwritten, see the comment in the overwritten method. additionally it _completely_ changes the meaning of ==, definitely making it unusable.
- <, > : one normally expects a Boolean as a return, not an array, and it is often not mappable in good math systems if i remember correctly. i wonder whether that is really necessary.

well ok, that was PMVector, i better stop now <friendly grin>.
werner

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

SergeStinckwich

Re: first impressions

In reply to this post by werner kassens-2

I'm not sure there are projects depending on PolyMath ?
I will just rename the method to asPMVector.

On Sat, Mar 18, 2017 at 10:49 AM, werner kassens <[hidden email]> wrote:

> yes Uko, i agree it is a popular method, you are probably right.
> werner
>
> On Sat, Mar 18, 2017 at 10:32 AM, Yuriy Tymchuk <[hidden email]> wrote:
>>
>> Yes, I understand, but I’m thinking about the projects that depend on
>> Polymath and I suppose #asDhbVector is a popular method there.
>>
>> Cheers
>> Uko
>>
>>
>> On 18 Mar 2017, at 10:22, werner kassens <[hidden email]> wrote:
>>
>> Hi Uko,
>> i guess a simple renaming would be sufficient, since #asDhbVector has
>> already been replaced by #asPMVector in most cases, there are only two
>> places where this has been forgotten.
>> werner
>>
>> On Fri, Mar 17, 2017 at 7:09 PM, Yuriy Tymchuk <[hidden email]> wrote:
>>>
>>> I think we should deprecate this one with an automatic rewrite.
>>>
>>> Uko
>>>
>>> On 17 Mar 2017, at 19:06, werner kassens <[hidden email]> wrote:
>>>
>>> #asDHBVector should perhaps be renamed?
>>>
>>>
>>>
>>> --
>>> You received this message because you are subscribed to the Google Groups
>>> "SciSmalltalk" group.
>>> To unsubscribe from this group and stop receiving emails from it, send an
>>> email to [hidden email].
>>> For more options, visit https://groups.google.com/d/optout.
>>
>>
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "SciSmalltalk" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to [hidden email].
>> For more options, visit https://groups.google.com/d/optout.
>>
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "SciSmalltalk" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to [hidden email].
>> For more options, visit https://groups.google.com/d/optout.
>
>
> --
> You received this message because you are subscribed to the Google Groups
> "SciSmalltalk" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [hidden email].
> For more options, visit https://groups.google.com/d/optout.

--
Serge Stinckwich
UCN & UMI UMMISCO 209 (IRD/UPMC)
Every DSL ends up being Smalltalk
http://www.doesnotunderstand.org/

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

SergeStinckwich

Re: first impressions

In reply to this post by werner kassens-2

On Fri, Mar 17, 2017 at 7:06 PM, werner kassens <[hidden email]> wrote:
> Hi,

Hi Werner,

> fwiw i had a look at PMVector:

BTW all these methods were added by one student of Uko.
So constructive comments are welcome, to help him wrote better code. I
ask him to do : https://github.com/PolyMathOrg/PolyMath/issues/19
I hope Orest is in mailing-list.

> - asDHBVector : in collection this is renamed to #asPMVector. but this does
> not exist in PMVector, hence it is not polymorphic any more. #asDHBVector
> should perhaps be renamed?

Yes we forget about this one when the rename all the classes I guess.
I rename the method.

> - sin : superfluous since PMVector is an Array that understands #sin.
> further it changes the usual meaning of #sin insofar as it changes the
> original object instead of returning a new object ( see the difference
> between #negated and #negate). this is _not_ how pharo deals with functions
> like that and opens up the possibility of a heck of new bugs.
> - cosh, log, sin, sinh, sqrt, tan, tanh: same problem as with sin.

Yes I remove all of them.
I just only keep cosh, sinh, tanh at the moment that was not implement
on Collection.
I will move these methods at the level of Collection like the others one.

> - cumsum : should also return a new object.btw this behaviour is in all (?)
> cases automatically transferred to matrices which obviously can also produce
> bugs there.
> c:=PMSymmetricMatrix rows:#((1 2)(2 1)).
> a:=c cumsum.
> c isSymmetric . "true"
> c. "a PMVector(1 3)
> a PMVector(2 3)"
> just a simple example why this general approach in all these functions is
> problematic.

I reimplement cumsum like :

cumsum
"Cumulative sum
#(1 2 3 4 5) cumsum = #(1 3 6 10 15)
"
| sum |
sum := 0.
^ self collect: [ :v | sum := sum + v. sum ]

I see that Roassal reimplement this method in SequenceableCollection like that.
Maybe we should push this kind of methods to Pharo ?

I also reimplement PMMatrix>>cumsum as :
cumsum
"Computes the cumulative sum for each row."

^ PMMatrix rows: (rows collect: [ :each | each cumsum ])

> - sum: is also implemented like sin and superfluous, but here the problem is
> even worse because the implementation is changed, and not really for the
> better.

Remove.

> - dot: : #dot: is used in floatarray and means - as expected - "dot product"
> there. it means something completely else here. you might want to look for
> another name.

what name do you propose ?

> - == : should _not_ be overwritten, see the comment in the overwritten
> method. additionally it _completely_ changes the meaning of ==, definitely
> making it unusable.

Yes sure, you are right. I remove this method.
== was also implemented on PMMatrix.

> - <, > : one normally expects a Boolean as a return, not an array, and it is
> often not mappable in good math systems if i remember correctly. i wonder
> whether that is really necessary.

Yeah I guess Orest try to mimics the behavior of Numpy, where you can
do something like that on vectors.
Do you have a better solution ?

> well ok, that was PMVector, i better stop now <friendly grin>.

I release 0.85.

Thank you.

--
Serge Stinckwich
UCN & UMI UMMISCO 209 (IRD/UPMC)
Every DSL ends up being Smalltalk
http://www.doesnotunderstand.org/

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

werner kassens-2

Re: first impressions

Hi Serge,
thank you for all this info, very friendly. i'm glad that you deleted those functions. regarding the functions you kept, they are part of numbersextensions. it would perhaps make more sense to implement these methods there (also to get rid of stupid dependencies) on a lower level collection, have a look at eg Collection *Collections-arithmetic & *Collections-arithmetic-collectors.

>I see that Roassal reimplement this method in SequenceableCollection like that.
Maybe we should push this kind of methods to Pharo ?

yes indeed.

>dot:...what name do you propose ?
no idea, multiply:, multiplyWith:, elementwiseMultiply: ?
btw the implementation is perhaps a bit problematic (think of vectors of different size).

>Yeah I guess Orest try to mimics the behavior of Numpy, where you can
do something like that on vectors.
Do you have a better solution ?
no. i really dont see why it should be such problem to write a oneliner like
a:=aVector collect:[:x|x<3]
if one needs it.

in the smalltalkhub versions i think that there are again a lot of objects & methods that got lost. essentialy unusable. i guess that doesnt matter to you, since you want to move to git anyway and there it obviously didnt happen. i just mention this since it would not make sense to update the catalog browser as long as it cant load git-projects. and if you announce a new version you might want to take that into account.
werner

SergeStinckwich

Re: first impressions

On Sun, Mar 19, 2017 at 3:59 PM, werner kassens <[hidden email]> wrote:
> Hi Serge,
> thank you for all this info, very friendly. i'm glad that you deleted those
> functions. regarding the functions you kept, they are part of
> numbersextensions. it would perhaps make more sense to implement these
> methods there (also to get rid of stupid dependencies) on a lower level
> collection, have a look at eg Collection *Collections-arithmetic &
> *Collections-arithmetic-collectors.

yes sure. Can you add an issue on issue tracker in order to remember
about this ?

>>I see that Roassal reimplement this method in SequenceableCollection like
>> that.
> Maybe we should push this kind of methods to Pharo ?
>
> yes indeed.
>
>>dot:...what name do you propose ?
> no idea, multiply:, multiplyWith:, elementwiseMultiply: ?
> btw the implementation is perhaps a bit problematic (think of vectors of
> different size).

Same, add an issue, so we remember.

>>Yeah I guess Orest try to mimics the behavior of Numpy, where you can
> do something like that on vectors.
> Do you have a better solution ?
> no. i really dont see why it should be such problem to write a oneliner like
> a:=aVector collect:[:x|x<3]
> if one needs it.
>
> in the smalltalkhub versions i think that there are again a lot of objects &
> methods that got lost. essentialy unusable. i guess that doesnt matter to
> you, since you want to move to git anyway and there it obviously didnt
> happen. i just mention this since it would not make sense to update the
> catalog browser as long as it cant load git-projects. and if you announce a
> new version you might want to take that into account.

I don't understand, which methods ? objects are lost ?

Regards,

--
Serge Stinckwich
UCN & UMI UMMISCO 209 (IRD/UPMC)
Every DSL ends up being Smalltalk
http://www.doesnotunderstand.org/

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

werner kassens-2

Re: first impressions

Hi Serge,

don't you remember, it has already happened in the earlier times? just load the last config via smalltalkhub in a pristine image and then do a config loadDevelopment. and then run a few tests. it is not really important which one, a lot are broke.

werner

On Sun, Mar 19, 2017 at 4:03 PM, Serge Stinckwich <[hidden email]> wrote:

On Sun, Mar 19, 2017 at 3:59 PM, werner kassens <[hidden email]> wrote:
> Hi Serge,
> thank you for all this info, very friendly. i'm glad that you deleted those
> functions. regarding the functions you kept, they are part of
> numbersextensions. it would perhaps make more sense to implement these
> methods there (also to get rid of stupid dependencies) on a lower level
> collection, have a look at eg Collection *Collections-arithmetic &
> *Collections-arithmetic-collectors.

yes sure. Can you add an issue on issue tracker in order to remember
about this ?

>>I see that Roassal reimplement this method in SequenceableCollection like
>> that.
> Maybe we should push this kind of methods to Pharo ?
>
> yes indeed.
>
>>dot:...what name do you propose ?
> no idea, multiply:, multiplyWith:, elementwiseMultiply: ?
> btw the implementation is perhaps a bit problematic (think of vectors of
> different size).

Same, add an issue, so we remember.

>>Yeah I guess Orest try to mimics the behavior of Numpy, where you can
> do something like that on vectors.
> Do you have a better solution ?
> no. i really dont see why it should be such problem to write a oneliner like
> a:=aVector collect:[:x|x<3]
> if one needs it.
>
> in the smalltalkhub versions i think that there are again a lot of objects &
> methods that got lost. essentialy unusable. i guess that doesnt matter to
> you, since you want to move to git anyway and there it obviously didnt
> happen. i just mention this since it would not make sense to update the
> catalog browser as long as it cant load git-projects. and if you announce a
> new version you might want to take that into account.

I don't understand, which methods ? objects are lost ?

Regards,

--
Serge Stinckwich
UCN & UMI UMMISCO 209 (IRD/UPMC)
Every DSL ends up being Smalltalk
http://www.doesnotunderstand.org/

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

werner kassens-2

Re: first impressions

or the second to last config which probably is the correct one as the last has a CI ending.

On Sun, Mar 19, 2017 at 4:10 PM, werner kassens <[hidden email]> wrote:

Hi Serge,
don't you remember, it has already happened in the earlier times? just load the last config via smalltalkhub in a pristine image and then do a config loadDevelopment. and then run a few tests. it is not really important which one, a lot are broke.
werner

On Sun, Mar 19, 2017 at 4:03 PM, Serge Stinckwich <[hidden email]> wrote:
On Sun, Mar 19, 2017 at 3:59 PM, werner kassens <[hidden email]> wrote:
> Hi Serge,
> thank you for all this info, very friendly. i'm glad that you deleted those
> functions. regarding the functions you kept, they are part of
> numbersextensions. it would perhaps make more sense to implement these
> methods there (also to get rid of stupid dependencies) on a lower level
> collection, have a look at eg Collection *Collections-arithmetic &
> *Collections-arithmetic-collectors.

yes sure. Can you add an issue on issue tracker in order to remember
about this ?

>>I see that Roassal reimplement this method in SequenceableCollection like
>> that.
> Maybe we should push this kind of methods to Pharo ?
>
> yes indeed.
>
>>dot:...what name do you propose ?
> no idea, multiply:, multiplyWith:, elementwiseMultiply: ?
> btw the implementation is perhaps a bit problematic (think of vectors of
> different size).

Same, add an issue, so we remember.

>>Yeah I guess Orest try to mimics the behavior of Numpy, where you can
> do something like that on vectors.
> Do you have a better solution ?
> no. i really dont see why it should be such problem to write a oneliner like
> a:=aVector collect:[:x|x<3]
> if one needs it.
>
> in the smalltalkhub versions i think that there are again a lot of objects &
> methods that got lost. essentialy unusable. i guess that doesnt matter to
> you, since you want to move to git anyway and there it obviously didnt
> happen. i just mention this since it would not make sense to update the
> catalog browser as long as it cant load git-projects. and if you announce a
> new version you might want to take that into account.

I don't understand, which methods ? objects are lost ?

Regards,

--
Serge Stinckwich
UCN & UMI UMMISCO 209 (IRD/UPMC)
Every DSL ends up being Smalltalk
http://www.doesnotunderstand.org/

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

SergeStinckwich

Re: first impressions

In reply to this post by werner kassens-2

On Sun, Mar 19, 2017 at 4:10 PM, werner kassens <[hidden email]> wrote:
> Hi Serge,
> don't you remember, it has already happened in the earlier times? just load
> the last config via smalltalkhub in a pristine image and then do a config
> loadDevelopment. and then run a few tests. it is not really important which
> one, a lot are broke.

I don't understand sorry. You mean tests are broken ?
Normally, PolyMath is loaded as :

Gofer new
url: 'http://smalltalkhub.com/mc/PolyMath/PolyMath/main';
configurationOf:'PolyMath';
loadVersion: '0.85'

in a pristine Pharo 5.0/6.0 image
All tests are green.

--
Serge Stinckwich
UCN & UMI UMMISCO 209 (IRD/UPMC)
Every DSL ends up being Smalltalk
http://www.doesnotunderstand.org/

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

werner kassens-2

Re: first impressions

ah, then it was a false alarm, sorry, glad to hear that!

werner

On Sun, Mar 19, 2017 at 4:18 PM, Serge Stinckwich <[hidden email]> wrote:

On Sun, Mar 19, 2017 at 4:10 PM, werner kassens <[hidden email]> wrote:
> Hi Serge,
> don't you remember, it has already happened in the earlier times? just load
> the last config via smalltalkhub in a pristine image and then do a config
> loadDevelopment. and then run a few tests. it is not really important which
> one, a lot are broke.

I don't understand sorry. You mean tests are broken ?
Normally, PolyMath is loaded as :

Gofer new
url: 'http://smalltalkhub.com/mc/PolyMath/PolyMath/main';
configurationOf:'PolyMath';
loadVersion: '0.85'

in a pristine Pharo 5.0/6.0 image
All tests are green.

--
Serge Stinckwich
UCN & UMI UMMISCO 209 (IRD/UPMC)
Every DSL ends up being Smalltalk
http://www.doesnotunderstand.org/

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

werner kassens-2

Re: first impressions

just wondering why is https://ci.inria.fr/pharo-contribution/job/PolyMath/ red?

werner

On Sun, Mar 19, 2017 at 4:21 PM, werner kassens <[hidden email]> wrote:

ah, then it was a false alarm, sorry, glad to hear that!
werner

On Sun, Mar 19, 2017 at 4:18 PM, Serge Stinckwich <[hidden email]> wrote:
On Sun, Mar 19, 2017 at 4:10 PM, werner kassens <[hidden email]> wrote:
> Hi Serge,
> don't you remember, it has already happened in the earlier times? just load
> the last config via smalltalkhub in a pristine image and then do a config
> loadDevelopment. and then run a few tests. it is not really important which
> one, a lot are broke.

I don't understand sorry. You mean tests are broken ?
Normally, PolyMath is loaded as :

Gofer new
url: 'http://smalltalkhub.com/mc/PolyMath/PolyMath/main';
configurationOf:'PolyMath';
loadVersion: '0.85'

in a pristine Pharo 5.0/6.0 image
All tests are green.

--
Serge Stinckwich
UCN & UMI UMMISCO 209 (IRD/UPMC)
Every DSL ends up being Smalltalk
http://www.doesnotunderstand.org/

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

SergeStinckwich

Re: first impressions

I'm not using anymore the CI provided by INRIA. I should stop it I guess.
All CI jobs are run on travis and appveyor now thanks to SmalltalkCI :

https://travis-ci.org/PolyMathOrg/PolyMath
https://ci.appveyor.com/project/SergeStinckwich/polymath

On Sun, Mar 19, 2017 at 4:26 PM, werner kassens <[hidden email]> wrote:

> just wondering why is https://ci.inria.fr/pharo-contribution/job/PolyMath/
> red?
> werner
>
> On Sun, Mar 19, 2017 at 4:21 PM, werner kassens <[hidden email]>
> wrote:
>>
>> ah, then it was a false alarm, sorry, glad to hear that!
>> werner
>>
>> On Sun, Mar 19, 2017 at 4:18 PM, Serge Stinckwich
>> <[hidden email]> wrote:
>>>
>>> On Sun, Mar 19, 2017 at 4:10 PM, werner kassens <[hidden email]>
>>> wrote:
>>> > Hi Serge,
>>> > don't you remember, it has already happened in the earlier times? just
>>> > load
>>> > the last config via smalltalkhub in a pristine image and then do a
>>> > config
>>> > loadDevelopment. and then run a few tests. it is not really important
>>> > which
>>> > one, a lot are broke.
>>>
>>> I don't understand sorry. You mean tests are broken ?
>>> Normally, PolyMath is loaded as :
>>>
>>> Gofer new
>>> url: 'http://smalltalkhub.com/mc/PolyMath/PolyMath/main';
>>> configurationOf:'PolyMath';
>>> loadVersion: '0.85'
>>>
>>> in a pristine Pharo 5.0/6.0 image
>>> All tests are green.
>>>
>>> --
>>> Serge Stinckwich
>>> UCN & UMI UMMISCO 209 (IRD/UPMC)
>>> Every DSL ends up being Smalltalk
>>> http://www.doesnotunderstand.org/
>>>
>>> --
>>> You received this message because you are subscribed to the Google Groups
>>> "SciSmalltalk" group.
>>> To unsubscribe from this group and stop receiving emails from it, send an
>>> email to [hidden email].
>>> For more options, visit https://groups.google.com/d/optout.
>>
>>
>
> --
> You received this message because you are subscribed to the Google Groups
> "SciSmalltalk" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [hidden email].
> For more options, visit https://groups.google.com/d/optout.

werner kassens-2

Re: first impressions

ah ok, i see.

werner

On Sun, Mar 19, 2017 at 4:30 PM, Serge Stinckwich <[hidden email]> wrote:

I'm not using anymore the CI provided by INRIA. I should stop it I guess.
All CI jobs are run on travis and appveyor now thanks to SmalltalkCI :

https://travis-ci.org/PolyMathOrg/PolyMath
https://ci.appveyor.com/project/SergeStinckwich/polymath

On Sun, Mar 19, 2017 at 4:26 PM, werner kassens <[hidden email]> wrote:
> just wondering why is https://ci.inria.fr/pharo-contribution/job/PolyMath/
> red?
> werner
>
> On Sun, Mar 19, 2017 at 4:21 PM, werner kassens <[hidden email]>
> wrote:
>>
>> ah, then it was a false alarm, sorry, glad to hear that!
>> werner
>>
>> On Sun, Mar 19, 2017 at 4:18 PM, Serge Stinckwich
>> <[hidden email]> wrote:
>>>
>>> On Sun, Mar 19, 2017 at 4:10 PM, werner kassens <[hidden email]>
>>> wrote:
>>> > Hi Serge,
>>> > don't you remember, it has already happened in the earlier times? just
>>> > load
>>> > the last config via smalltalkhub in a pristine image and then do a
>>> > config
>>> > loadDevelopment. and then run a few tests. it is not really important
>>> > which
>>> > one, a lot are broke.
>>>
>>> I don't understand sorry. You mean tests are broken ?
>>> Normally, PolyMath is loaded as :
>>>
>>> Gofer new
>>> url: 'http://smalltalkhub.com/mc/PolyMath/PolyMath/main';
>>> configurationOf:'PolyMath';
>>> loadVersion: '0.85'
>>>
>>> in a pristine Pharo 5.0/6.0 image
>>> All tests are green.
>>>
>>> --
>>> Serge Stinckwich
>>> UCN & UMI UMMISCO 209 (IRD/UPMC)
>>> Every DSL ends up being Smalltalk
>>> http://www.doesnotunderstand.org/
>>>
>>> --
>>> You received this message because you are subscribed to the Google Groups
>>> "SciSmalltalk" group.
>>> To unsubscribe from this group and stop receiving emails from it, send an
>>> email to [hidden email].
>>> For more options, visit https://groups.google.com/d/optout.
>>
>>
>
> --
> You received this message because you are subscribed to the Google Groups
> "SciSmalltalk" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [hidden email].
> For more options, visit https://groups.google.com/d/optout.

--
Serge Stinckwich
UCN & UMI UMMISCO 209 (IRD/UPMC)
Every DSL ends up being Smalltalk
http://www.doesnotunderstand.org/

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

SergeStinckwich

Re: first impressions

In reply to this post by SergeStinckwich

I remove the Jenkins job on inria CI in order to avoid confusion.

Regards,

On Sun, Mar 19, 2017 at 4:30 PM, Serge Stinckwich
<[hidden email]> wrote:

> I'm not using anymore the CI provided by INRIA. I should stop it I guess.
> All CI jobs are run on travis and appveyor now thanks to SmalltalkCI :
>
> https://travis-ci.org/PolyMathOrg/PolyMath
> https://ci.appveyor.com/project/SergeStinckwich/polymath
>
>
>
> On Sun, Mar 19, 2017 at 4:26 PM, werner kassens <[hidden email]> wrote:
>> just wondering why is https://ci.inria.fr/pharo-contribution/job/PolyMath/
>> red?
>> werner
>>
>> On Sun, Mar 19, 2017 at 4:21 PM, werner kassens <[hidden email]>
>> wrote:
>>>
>>> ah, then it was a false alarm, sorry, glad to hear that!
>>> werner
>>>
>>> On Sun, Mar 19, 2017 at 4:18 PM, Serge Stinckwich
>>> <[hidden email]> wrote:
>>>>
>>>> On Sun, Mar 19, 2017 at 4:10 PM, werner kassens <[hidden email]>
>>>> wrote:
>>>> > Hi Serge,
>>>> > don't you remember, it has already happened in the earlier times? just
>>>> > load
>>>> > the last config via smalltalkhub in a pristine image and then do a
>>>> > config
>>>> > loadDevelopment. and then run a few tests. it is not really important
>>>> > which
>>>> > one, a lot are broke.
>>>>
>>>> I don't understand sorry. You mean tests are broken ?
>>>> Normally, PolyMath is loaded as :
>>>>
>>>> Gofer new
>>>> url: 'http://smalltalkhub.com/mc/PolyMath/PolyMath/main';
>>>> configurationOf:'PolyMath';
>>>> loadVersion: '0.85'
>>>>
>>>> in a pristine Pharo 5.0/6.0 image
>>>> All tests are green.
>>>>
>>>> --
>>>> Serge Stinckwich
>>>> UCN & UMI UMMISCO 209 (IRD/UPMC)
>>>> Every DSL ends up being Smalltalk
>>>> http://www.doesnotunderstand.org/
>>>>
>>>> --
>>>> You received this message because you are subscribed to the Google Groups
>>>> "SciSmalltalk" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send an
>>>> email to [hidden email].
>>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>>
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "SciSmalltalk" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to [hidden email].
>> For more options, visit https://groups.google.com/d/optout.
>
>
>
> --
> Serge Stinckwich
> UCN & UMI UMMISCO 209 (IRD/UPMC)
> Every DSL ends up being Smalltalk
> http://www.doesnotunderstand.org/

werner kassens-2

Re: first impressions

Hi Serge,
sorry for bothering you, i did a
Gofer new
    url: 'http://smalltalkhub.com/mc/PolyMath/PolyMath/main';
    configurationOf:'PolyMath';
    loadVersion: '0.85'
in an unused pharo-50770.image, which should be the latest 5.0 one, and got an error
"The symbolic version #'0.85' is not defined in ConfigurationOfPolyMath for the current platform. Possible symbolic version values include: #(#bleedingEdge #development #stable)"
i then did
Gofer new
    url: 'http://smalltalkhub.com/mc/PolyMath/PolyMath/main';
    configurationOf:'PolyMath';
    loadVersion: '0.40-baseline'
which worked, but in the tests i get a load of errors eg in Math-Tests-AutomaticDifferenciation. as i said, in a new 5.0 image. and for you everything works in 5.0 with polymath loaded from smalltalkhub?
werner

On Sun, Mar 19, 2017 at 5:58 PM, Serge Stinckwich <[hidden email]> wrote:

I remove the Jenkins job on inria CI in order to avoid confusion.

Regards,

On Sun, Mar 19, 2017 at 4:30 PM, Serge Stinckwich

<[hidden email]> wrote:
> I'm not using anymore the CI provided by INRIA. I should stop it I guess.
> All CI jobs are run on travis and appveyor now thanks to SmalltalkCI :
>
> https://travis-ci.org/PolyMathOrg/PolyMath
> https://ci.appveyor.com/project/SergeStinckwich/polymath
>
>
>
> On Sun, Mar 19, 2017 at 4:26 PM, werner kassens <[hidden email]> wrote:
>> just wondering why is https://ci.inria.fr/pharo-contribution/job/PolyMath/
>> red?
>> werner
>>
>> On Sun, Mar 19, 2017 at 4:21 PM, werner kassens <[hidden email]>
>> wrote:
>>>
>>> ah, then it was a false alarm, sorry, glad to hear that!
>>> werner
>>>
>>> On Sun, Mar 19, 2017 at 4:18 PM, Serge Stinckwich
>>> <[hidden email]> wrote:
>>>>
>>>> On Sun, Mar 19, 2017 at 4:10 PM, werner kassens <[hidden email]>
>>>> wrote:
>>>> > Hi Serge,
>>>> > don't you remember, it has already happened in the earlier times? just
>>>> > load
>>>> > the last config via smalltalkhub in a pristine image and then do a
>>>> > config
>>>> > loadDevelopment. and then run a few tests. it is not really important
>>>> > which
>>>> > one, a lot are broke.
>>>>
>>>> I don't understand sorry. You mean tests are broken ?
>>>> Normally, PolyMath is loaded as :
>>>>
>>>> Gofer new
>>>> url: 'http://smalltalkhub.com/mc/PolyMath/PolyMath/main';
>>>> configurationOf:'PolyMath';
>>>> loadVersion: '0.85'
>>>>
>>>> in a pristine Pharo 5.0/6.0 image
>>>> All tests are green.
>>>>
>>>> --
>>>> Serge Stinckwich
>>>> UCN & UMI UMMISCO 209 (IRD/UPMC)
>>>> Every DSL ends up being Smalltalk
>>>> http://www.doesnotunderstand.org/
>>>>
>>>> --
>>>> You received this message because you are subscribed to the Google Groups
>>>> "SciSmalltalk" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send an
>>>> email to [hidden email].
>>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>>
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "SciSmalltalk" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to [hidden email].
>> For more options, visit https://groups.google.com/d/optout.
>
>
>
> --
> Serge Stinckwich
> UCN & UMI UMMISCO 209 (IRD/UPMC)
> Every DSL ends up being Smalltalk
> http://www.doesnotunderstand.org/

--
Serge Stinckwich
UCN & UMI UMMISCO 209 (IRD/UPMC)
Every DSL ends up being Smalltalk
http://www.doesnotunderstand.org/

--
You received this message because you are subscribed to the Google Groups "SciSmalltalk" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

SergeStinckwich

Re: first impressions

I push a new version, this should works now:

Gofer new
url: 'http://smalltalkhub.com/mc/PolyMath/PolyMath/main';
configurationOf:'PolyMath';
loadVersion: '0.85'

On Sun, Mar 19, 2017 at 6:22 PM, werner kassens <[hidden email]> wrote:

> Hi Serge,
> sorry for bothering you, i did a
> Gofer new
> url: 'http://smalltalkhub.com/mc/PolyMath/PolyMath/main';
> configurationOf:'PolyMath';
> loadVersion: '0.85'
> in an unused pharo-50770.image, which should be the latest 5.0 one, and got
> an error
> "The symbolic version #'0.85' is not defined in ConfigurationOfPolyMath for
> the current platform. Possible symbolic version values include:
> #(#bleedingEdge #development #stable)"
> i then did
> Gofer new
> url: 'http://smalltalkhub.com/mc/PolyMath/PolyMath/main';
> configurationOf:'PolyMath';
> loadVersion: '0.40-baseline'
> which worked, but in the tests i get a load of errors eg in
> Math-Tests-AutomaticDifferenciation. as i said, in a new 5.0 image. and for
> you everything works in 5.0 with polymath loaded from smalltalkhub?
> werner
>
> On Sun, Mar 19, 2017 at 5:58 PM, Serge Stinckwich
> <[hidden email]> wrote:
>>
>> I remove the Jenkins job on inria CI in order to avoid confusion.
>>
>> Regards,
>>
>> On Sun, Mar 19, 2017 at 4:30 PM, Serge Stinckwich
>> <[hidden email]> wrote:
>> > I'm not using anymore the CI provided by INRIA. I should stop it I
>> > guess.
>> > All CI jobs are run on travis and appveyor now thanks to SmalltalkCI :
>> >
>> > https://travis-ci.org/PolyMathOrg/PolyMath
>> > https://ci.appveyor.com/project/SergeStinckwich/polymath
>> >
>> >
>> >
>> > On Sun, Mar 19, 2017 at 4:26 PM, werner kassens <[hidden email]>
>> > wrote:
>> >> just wondering why is
>> >> https://ci.inria.fr/pharo-contribution/job/PolyMath/
>> >> red?
>> >> werner
>> >>
>> >> On Sun, Mar 19, 2017 at 4:21 PM, werner kassens <[hidden email]>
>> >> wrote:
>> >>>
>> >>> ah, then it was a false alarm, sorry, glad to hear that!
>> >>> werner
>> >>>
>> >>> On Sun, Mar 19, 2017 at 4:18 PM, Serge Stinckwich
>> >>> <[hidden email]> wrote:
>> >>>>
>> >>>> On Sun, Mar 19, 2017 at 4:10 PM, werner kassens
>> >>>> <[hidden email]>
>> >>>> wrote:
>> >>>> > Hi Serge,
>> >>>> > don't you remember, it has already happened in the earlier times?
>> >>>> > just
>> >>>> > load
>> >>>> > the last config via smalltalkhub in a pristine image and then do a
>> >>>> > config
>> >>>> > loadDevelopment. and then run a few tests. it is not really
>> >>>> > important
>> >>>> > which
>> >>>> > one, a lot are broke.
>> >>>>
>> >>>> I don't understand sorry. You mean tests are broken ?
>> >>>> Normally, PolyMath is loaded as :
>> >>>>
>> >>>> Gofer new
>> >>>> url: 'http://smalltalkhub.com/mc/PolyMath/PolyMath/main';
>> >>>> configurationOf:'PolyMath';
>> >>>> loadVersion: '0.85'
>> >>>>
>> >>>> in a pristine Pharo 5.0/6.0 image
>> >>>> All tests are green.
>> >>>>
>> >>>> --
>> >>>> Serge Stinckwich
>> >>>> UCN & UMI UMMISCO 209 (IRD/UPMC)
>> >>>> Every DSL ends up being Smalltalk
>> >>>> http://www.doesnotunderstand.org/
>> >>>>
>> >>>> --
>> >>>> You received this message because you are subscribed to the Google
>> >>>> Groups
>> >>>> "SciSmalltalk" group.
>> >>>> To unsubscribe from this group and stop receiving emails from it,
>> >>>> send an
>> >>>> email to [hidden email].
>> >>>> For more options, visit https://groups.google.com/d/optout.
>> >>>
>> >>>
>> >>
>> >> --
>> >> You received this message because you are subscribed to the Google
>> >> Groups
>> >> "SciSmalltalk" group.
>> >> To unsubscribe from this group and stop receiving emails from it, send
>> >> an
>> >> email to [hidden email].
>> >> For more options, visit https://groups.google.com/d/optout.
>> >
>> >
>> >
>> > --
>> > Serge Stinckwich
>> > UCN & UMI UMMISCO 209 (IRD/UPMC)
>> > Every DSL ends up being Smalltalk
>> > http://www.doesnotunderstand.org/
>>
>>
>>
>> --
>> Serge Stinckwich
>> UCN & UMI UMMISCO 209 (IRD/UPMC)
>> Every DSL ends up being Smalltalk
>> http://www.doesnotunderstand.org/
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "SciSmalltalk" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to [hidden email].
>> For more options, visit https://groups.google.com/d/optout.
>
>
> --
> You received this message because you are subscribed to the Google Groups
> "SciSmalltalk" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [hidden email].
> For more options, visit https://groups.google.com/d/optout.