Looks like a great On Mon, Feb 16, 2015 at 3:48 AM, Sebastian Sastre < sebastian@flowingconcept.com> wrote:

Great!

Forked repo, starred and PR sent: https://github.com/pharo-project/pharo-project-proposals/pull/1

keep up the good work!

I see this is merged 19 hours ago, but it doesn't show on the page for me. e.g. "7GUIs" btw, * Perhaps pharo-users would be a less daunting mail list for students begin their enquiries on. * Any chance of having working urls so its easier to engage candidates with links to further info?

I am not an expert in Pharo in any conceivable way, but I have tried to learn it over the past few weeks. I think that an area where the interactive tools of Pharo would really shine is that of data visualization, inspection and mining. Pharo already has very good tools - such as Roassal2 - for data visualization, but I think it lacks support for machine learning and scientific computing in general. The most useful thing I would like to come to Pharo (apart from performance improvements in the VM itself) would be something like scikit-learn for python. Also, these tasks often involve consuming data from various sources, such as CSV and Json files. NeoCSV and NeoJSON are still a little too rigid for the task - libraries like pandas allow to just feed a csv file and try to make head or tails of the content without having to define too much of a schema beforehand The second thing I can think of is a modern, minimal web framework. I know of Seaside and Seaside REST, but it feels too big and complex. Most web projects as of today involve some kind of stateless API that is consumed by single page applications or mobile apps, and Seaside does not really feel like a good solution for these cases. I would like to see something much smaller, like Spray (scala) or Sinatra (ruby). 2015-02-17 19:28 GMT+01:00 Serge Stinckwich <serge.stinckwich@gmail.com>:

We have something like 45 projects ideas at the moment. We really need more project ideas from more people (not only RMOD guys).

Even if you have a vague idea, you can contribute.

Thank you.

Regards,

On Sun, Feb 15, 2015 at 3:23 PM, Serge Stinckwich <serge.stinckwich@gmail.com> wrote:

...

Dear pharoers,

this year Pharo consortium (and community) is going to take part in a Google Summer of Code event[1] as a standalone organization. This is an opportunity to promote Pharo, get some job done and have students paid.

Currently we are at the most important stage as we are preparing the organization application, and hoping that we will be accepted and granted decent amount of project slots. Everyone can help with application by submitting ideas for student projects.

Current list can be found at: https://github.com/pharo-project/pharo-project-proposals/blob/master/Topics....

It is in STON format, and result is being generated at: http://gsoc.pharo.org/

Please add your ideas following the format of existing projects and open a pull request with them (you will need a github account). Preferably submit ideas with possible mentors, but if none are available at the moment ideas without mentors are also welcome.

The template to submit projects is :

PharoTopic new title: 'The name of your project; contact: 'email address'; supervisors: 'Supervisors names'; keywords: 'keywords separated by spaces; context: 'a description of the context of the project'; goal: 'description of the goal'; level: 'Beginner or Intermediate or Advanced'; yourself.

We will need a lot of projects/idea before February 20th 2015, the deadline for applying to GSOC 2015.

Do not hesitate to ask questions. Administrators of this year’s application are Serge Stinckwich <serge.stinckwich@gmail.com> and Yuriy Tymchuk <yuriy.tymchuk@me.com>

If you don't know how to edit the list, please send your project following the template to the administrators.

[1]: https://www.google-melange.com/gsoc/homepage/google/gsoc2015

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

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

On 18 Feb 2015, at 09:52, Andrea Ferretti <ferrettiandrea@gmail.com> wrote:

Also, these tasks often involve consuming data from various sources, such as CSV and Json files. NeoCSV and NeoJSON are still a little too rigid for the task - libraries like pandas allow to just feed a csv file and try to make head or tails of the content without having to define too much of a schema beforehand

Both NeoCSV and NeoJSON can operate in two ways, (1) without the definition of any schema's or (2) with the definition of schema's and mappings. The quick and dirty explore style is most certainly possible. 'my-data.csv' asFileReference readStreamDo: [ :in | (NeoCSVReader on: in) upToEnd ]. => an array of arrays 'my-data.json' asFileReference readStreamDo: [ :in | (NeoJSONReader on: in) next ]. => objects structured using dictionaries and arrays Sven

For an example of what I am talking about, see http://pandas.pydata.org/pandas-docs/version/0.15.2/io.html#csv-text-files I agree that this is definitely too much options, but it gets the job done for quick and dirty exploration. The fact is that working with a dump of table on your db, whose content you know, requires different tools than exploring the latest opendata that your local municipality has put online, using yet another messy format. Enterprise programmers deal more often with the former, data scientists with the latter, and I think there is room for both kind of tools 2015-02-18 10:26 GMT+01:00 Andrea Ferretti <ferrettiandrea@gmail.com>:

Thank you Sven. I think this should be emphasized and prominent on the home page*. Still, libraries such as pandas are even more lenient, doing things such as:

- autodetecting which fields are numeric in CSV files - allowing to fill missing data based on statistics (for instance, you can say: where the field `age` is missing, use the average age)

Probably there is room for something built on top of Neo

* by the way, I suggest that the documentation on Neo could benefit from a reorganization. Right now, the first topic on the NeoJSON paper introduces JSON itself. I would argue that everyone that tries to use the library knows what JSON is already. Still, there is no example of how to read JSON from a file in the whole document.

2015-02-18 10:12 GMT+01:00 Sven Van Caekenberghe <sven@stfx.eu>:

...

...

On 18 Feb 2015, at 09:52, Andrea Ferretti <ferrettiandrea@gmail.com> wrote:

Also, these tasks often involve consuming data from various sources, such as CSV and Json files. NeoCSV and NeoJSON are still a little too rigid for the task - libraries like pandas allow to just feed a csv file and try to make head or tails of the content without having to define too much of a schema beforehand

Both NeoCSV and NeoJSON can operate in two ways, (1) without the definition of any schema's or (2) with the definition of schema's and mappings. The quick and dirty explore style is most certainly possible.

'my-data.csv' asFileReference readStreamDo: [ :in | (NeoCSVReader on: in) upToEnd ].

=> an array of arrays

'my-data.json' asFileReference readStreamDo: [ :in | (NeoJSONReader on: in) next ].

=> objects structured using dictionaries and arrays

Sven

I am sorry if the previous messages came off as too harsh. The Neo tools are perfectly fine for their intended use. What I was trying to say is that a good idea for a SoC project would be to develop a framework for data analysis that would be useful for data scientists, and in particular this would include something to import unstructured data more freely. 2015-02-18 10:39 GMT+01:00 Sven Van Caekenberghe <sven@stfx.eu>:

Well, you are certainly free to contribute.

Heuristic interpretation of data could be useful, but looks like an addition on top, the core library should be fast and efficient.

...

On 18 Feb 2015, at 10:35, Andrea Ferretti <ferrettiandrea@gmail.com> wrote:

For an example of what I am talking about, see

http://pandas.pydata.org/pandas-docs/version/0.15.2/io.html#csv-text-files

I agree that this is definitely too much options, but it gets the job done for quick and dirty exploration.

The fact is that working with a dump of table on your db, whose content you know, requires different tools than exploring the latest opendata that your local municipality has put online, using yet another messy format.

Enterprise programmers deal more often with the former, data scientists with the latter, and I think there is room for both kind of tools

2015-02-18 10:26 GMT+01:00 Andrea Ferretti <ferrettiandrea@gmail.com>:

...

Thank you Sven. I think this should be emphasized and prominent on the home page*. Still, libraries such as pandas are even more lenient, doing things such as:

- autodetecting which fields are numeric in CSV files - allowing to fill missing data based on statistics (for instance, you can say: where the field `age` is missing, use the average age)

Probably there is room for something built on top of Neo

* by the way, I suggest that the documentation on Neo could benefit from a reorganization. Right now, the first topic on the NeoJSON paper introduces JSON itself. I would argue that everyone that tries to use the library knows what JSON is already. Still, there is no example of how to read JSON from a file in the whole document.

2015-02-18 10:12 GMT+01:00 Sven Van Caekenberghe <sven@stfx.eu>:

...

...

On 18 Feb 2015, at 09:52, Andrea Ferretti <ferrettiandrea@gmail.com> wrote:

Also, these tasks often involve consuming data from various sources, such as CSV and Json files. NeoCSV and NeoJSON are still a little too rigid for the task - libraries like pandas allow to just feed a csv file and try to make head or tails of the content without having to define too much of a schema beforehand

Both NeoCSV and NeoJSON can operate in two ways, (1) without the definition of any schema's or (2) with the definition of schema's and mappings. The quick and dirty explore style is most certainly possible.

'my-data.csv' asFileReference readStreamDo: [ :in | (NeoCSVReader on: in) upToEnd ].

=> an array of arrays

'my-data.json' asFileReference readStreamDo: [ :in | (NeoJSONReader on: in) next ].

=> objects structured using dictionaries and arrays

Sven

I am sorry, I must have misunderstood the purpose of this thread. I read "Even if you have a vague idea, you can contribute." and tried to give a couple of vague ideas. I did not really mean that I would be able or have time to mentor such a project 2015-02-18 11:01 GMT+01:00 Sven Van Caekenberghe <sven@stfx.eu>:

OK, try making a proposal then, http://gsoc.pharo.org has the instructions and the current list, you probably know more about data science than I do.

...

On 18 Feb 2015, at 10:53, Andrea Ferretti <ferrettiandrea@gmail.com> wrote:

I am sorry if the previous messages came off as too harsh. The Neo tools are perfectly fine for their intended use.

What I was trying to say is that a good idea for a SoC project would be to develop a framework for data analysis that would be useful for data scientists, and in particular this would include something to import unstructured data more freely.

2015-02-18 10:39 GMT+01:00 Sven Van Caekenberghe <sven@stfx.eu>:

...

Well, you are certainly free to contribute.

Heuristic interpretation of data could be useful, but looks like an addition on top, the core library should be fast and efficient.

...

On 18 Feb 2015, at 10:35, Andrea Ferretti <ferrettiandrea@gmail.com> wrote:

For an example of what I am talking about, see

http://pandas.pydata.org/pandas-docs/version/0.15.2/io.html#csv-text-files

I agree that this is definitely too much options, but it gets the job done for quick and dirty exploration.

The fact is that working with a dump of table on your db, whose content you know, requires different tools than exploring the latest opendata that your local municipality has put online, using yet another messy format.

Enterprise programmers deal more often with the former, data scientists with the latter, and I think there is room for both kind of tools

2015-02-18 10:26 GMT+01:00 Andrea Ferretti <ferrettiandrea@gmail.com>:

...

Thank you Sven. I think this should be emphasized and prominent on the home page*. Still, libraries such as pandas are even more lenient, doing things such as:

- autodetecting which fields are numeric in CSV files - allowing to fill missing data based on statistics (for instance, you can say: where the field `age` is missing, use the average age)

Probably there is room for something built on top of Neo

* by the way, I suggest that the documentation on Neo could benefit from a reorganization. Right now, the first topic on the NeoJSON paper introduces JSON itself. I would argue that everyone that tries to use the library knows what JSON is already. Still, there is no example of how to read JSON from a file in the whole document.

2015-02-18 10:12 GMT+01:00 Sven Van Caekenberghe <sven@stfx.eu>:

...

...

On 18 Feb 2015, at 09:52, Andrea Ferretti <ferrettiandrea@gmail.com> wrote:

Also, these tasks often involve consuming data from various sources, such as CSV and Json files. NeoCSV and NeoJSON are still a little too rigid for the task - libraries like pandas allow to just feed a csv file and try to make head or tails of the content without having to define too much of a schema beforehand

Both NeoCSV and NeoJSON can operate in two ways, (1) without the definition of any schema's or (2) with the definition of schema's and mappings. The quick and dirty explore style is most certainly possible.

'my-data.csv' asFileReference readStreamDo: [ :in | (NeoCSVReader on: in) upToEnd ].

=> an array of arrays

'my-data.json' asFileReference readStreamDo: [ :in | (NeoJSONReader on: in) next ].

=> objects structured using dictionaries and arrays

Sven

On Wed, Feb 18, 2015 at 11:01 AM, Sven Van Caekenberghe <sven@stfx.eu> wrote:

OK, try making a proposal then, http://gsoc.pharo.org has the instructions and the current list, you probably know more about data science than I do.

...

On 18 Feb 2015, at 10:53, Andrea Ferretti <ferrettiandrea@gmail.com> wrote:

I am sorry if the previous messages came off as too harsh. The Neo tools are perfectly fine for their intended use.

What I was trying to say is that a good idea for a SoC project would be to develop a framework for data analysis that would be useful for data scientists, and in particular this would include something to import unstructured data more freely.

Sorry Andrea. I didn't see you message because I'm not pharo-users mailing-list, only on pharo-dev. I'm also really interested to have a gsoc project to develop data analysis framework. Please let's talk together in order to discuss about a proposal. Regards, -- Serge Stinckwich UCBN & UMI UMMISCO 209 (IRD/UPMC) Every DSL ends up being Smalltalk http://www.doesnotunderstand.org/

Indeed these are nice to have, now they will not magically happen :) There is a 400 pages book on SciTalk. Stef Le 19/2/15 09:36, Andrea Ferretti a écrit :

Hi Serge,

as I said I do not really have the time now to get involved in a gsoc proposal, but I can give you my perspective. There are two sides to the story.

The first one is complementary to SciSmalltalk: in order to analize data, you need to get data in first. So, one may want to read - say - a CSV, and have a number of heuristics, such as:

- autodetection of encoding - autodetection of quotes and delimiter - autodetection of columns containing numbers or dates - the possibility to indicate that some markers, such as "N/A", represent missing values - the possibility to indicate a replacement for missing values, such as 0, or "", or the average or the minimum of the other values in the colums

See http://pandas.pydata.org/pandas-docs/version/0.15.2/io.html#csv-text-files for some examples.

It may be worth to consider making this into a sequence that is read and processed lazily, to deal with CSV files bigger than memory.

When data is finally in, usually the first task is doing some processing, inspection or visualization. The Smalltalk collections are good for processing (although some lazy variants might help), and Roassal and the inspectors are perfect for visualization and browsing.

The second part comes the time when one wants to run some algorithm. While there is no need to have the fanciest ones, there should be some of the basics, such as:

- some form or regression (linear, logistic...) - some form of clustering (kmeans, dbscan, canopy...) - SVM

Another thing which would be useful is support for linear algebra, leveraging native libraries such as BLAS or LAPACK.

In short: just copying R, or numpy + pandas + scikit-learn would already be a giant leap forward.

Actually, some of the things I have mentioned above are already (I think) in SciSmalltalk, which brings me to the next point: documentation. There is really no point in having all these tools if people do not know they are there.

For this to become useful, there should be a dedicated site, highlighting what is already available, in what state (experimental, partial, stable...) and how to use it.

Ideally, I would include also some tutorials, for instance for dealing with standard problems such as Kaggle competitions. Here I think Smalltalk would have an edge, since these tutorial could be in the form of Prof Stef. Still, it would be nice if some form of the tutorials was also on the web, which makes it discoverable.

Best, Andrea

2015-02-18 11:14 GMT+01:00 Serge Stinckwich <serge.stinckwich@gmail.com>:

...

On Wed, Feb 18, 2015 at 11:01 AM, Sven Van Caekenberghe <sven@stfx.eu> wrote:

...

OK, try making a proposal then, http://gsoc.pharo.org has the instructions and the current list, you probably know more about data science than I do.

...

On 18 Feb 2015, at 10:53, Andrea Ferretti <ferrettiandrea@gmail.com> wrote:

I am sorry if the previous messages came off as too harsh. The Neo tools are perfectly fine for their intended use.

What I was trying to say is that a good idea for a SoC project would be to develop a framework for data analysis that would be useful for data scientists, and in particular this would include something to import unstructured data more freely. Sorry Andrea. I didn't see you message because I'm not pharo-users mailing-list, only on pharo-dev. I'm also really interested to have a gsoc project to develop data analysis framework. Please let's talk together in order to discuss about a proposal.

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

I will add these ideas to the list of Pharo topics (if serge did not do it yet) but after the result of GSOC are announced to avoid to break the web site. Stef Le 19/2/15 09:36, Andrea Ferretti a écrit :

Hi Serge,

as I said I do not really have the time now to get involved in a gsoc proposal, but I can give you my perspective. There are two sides to the story.

The first one is complementary to SciSmalltalk: in order to analize data, you need to get data in first. So, one may want to read - say - a CSV, and have a number of heuristics, such as:

- autodetection of encoding - autodetection of quotes and delimiter - autodetection of columns containing numbers or dates - the possibility to indicate that some markers, such as "N/A", represent missing values - the possibility to indicate a replacement for missing values, such as 0, or "", or the average or the minimum of the other values in the colums

See http://pandas.pydata.org/pandas-docs/version/0.15.2/io.html#csv-text-files for some examples.

It may be worth to consider making this into a sequence that is read and processed lazily, to deal with CSV files bigger than memory.

When data is finally in, usually the first task is doing some processing, inspection or visualization. The Smalltalk collections are good for processing (although some lazy variants might help), and Roassal and the inspectors are perfect for visualization and browsing.

The second part comes the time when one wants to run some algorithm. While there is no need to have the fanciest ones, there should be some of the basics, such as:

- some form or regression (linear, logistic...) - some form of clustering (kmeans, dbscan, canopy...) - SVM

Another thing which would be useful is support for linear algebra, leveraging native libraries such as BLAS or LAPACK.

In short: just copying R, or numpy + pandas + scikit-learn would already be a giant leap forward.

Actually, some of the things I have mentioned above are already (I think) in SciSmalltalk, which brings me to the next point: documentation. There is really no point in having all these tools if people do not know they are there.

For this to become useful, there should be a dedicated site, highlighting what is already available, in what state (experimental, partial, stable...) and how to use it.

Ideally, I would include also some tutorials, for instance for dealing with standard problems such as Kaggle competitions. Here I think Smalltalk would have an edge, since these tutorial could be in the form of Prof Stef. Still, it would be nice if some form of the tutorials was also on the web, which makes it discoverable.

Best, Andrea

2015-02-18 11:14 GMT+01:00 Serge Stinckwich <serge.stinckwich@gmail.com>:

...

On Wed, Feb 18, 2015 at 11:01 AM, Sven Van Caekenberghe <sven@stfx.eu> wrote:

...

OK, try making a proposal then, http://gsoc.pharo.org has the instructions and the current list, you probably know more about data science than I do.

...

On 18 Feb 2015, at 10:53, Andrea Ferretti <ferrettiandrea@gmail.com> wrote:

I am sorry if the previous messages came off as too harsh. The Neo tools are perfectly fine for their intended use.

What I was trying to say is that a good idea for a SoC project would be to develop a framework for data analysis that would be useful for data scientists, and in particular this would include something to import unstructured data more freely. Sorry Andrea. I didn't see you message because I'm not pharo-users mailing-list, only on pharo-dev. I'm also really interested to have a gsoc project to develop data analysis framework. Please let's talk together in order to discuss about a proposal.

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

On 18 Feb 2015, at 10:26, Andrea Ferretti <ferrettiandrea@gmail.com> wrote:

Thank you Sven. I think this should be emphasized and prominent on the home page*. Still, libraries such as pandas are even more lenient, doing things such as:

- autodetecting which fields are numeric in CSV files - allowing to fill missing data based on statistics (for instance, you can say: where the field `age` is missing, use the average age)

Probably there is room for something built on top of Neo

* by the way, I suggest that the documentation on Neo could benefit from a reorganization. Right now, the first topic on the NeoJSON paper introduces JSON itself. I would argue that everyone that tries to use the library knows what JSON is already. Still, there is no example of how to read JSON from a file in the whole document.

These libraries (NeoCSV, NeoJSON, STON) were all written with only a dependency on a limited character stream API. It was a design decision not to depend on a File API, because at the time we were transitioning from the old FileStreams to FileSystem. And I disagree about the JSON introduction ;-) You might know it, but that is not the case for everyone. Like not everyone knows CSV, HTTP, ... But I do agree that sometimes I too would like a convenience method here or there ;-)

2015-02-18 10:12 GMT+01:00 Sven Van Caekenberghe <sven@stfx.eu>:

...

...

On 18 Feb 2015, at 09:52, Andrea Ferretti <ferrettiandrea@gmail.com> wrote:

Also, these tasks often involve consuming data from various sources, such as CSV and Json files. NeoCSV and NeoJSON are still a little too rigid for the task - libraries like pandas allow to just feed a csv file and try to make head or tails of the content without having to define too much of a schema beforehand

Both NeoCSV and NeoJSON can operate in two ways, (1) without the definition of any schema's or (2) with the definition of schema's and mappings. The quick and dirty explore style is most certainly possible.

'my-data.csv' asFileReference readStreamDo: [ :in | (NeoCSVReader on: in) upToEnd ].

=> an array of arrays

'my-data.json' asFileReference readStreamDo: [ :in | (NeoJSONReader on: in) next ].

=> objects structured using dictionaries and arrays

Sven