Smalltalk › Pharo › Pharo Smalltalk Users

Basic versioning of GitHub repositories from within Pharo

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

35 messages Options

abergel

Re: Basic versioning of GitHub repositories from within Pharo

Yes, keep going

Alexandre

> Le 23 nov. 2015 à 15:05, Sven Van Caekenberghe <[hidden email]> a écrit :
>
> Yes, this is a cool project/approach.
>
> I personally love it that it does not require a plugin and/or library that might not exist on all platforms.
>
> Please keep going !
>
> Sven
>
>> On 23 Nov 2015, at 18:32, Thierry Goubier <[hidden email]> wrote:
>>
>> Hi Skip,
>>
>> this is interesting. It would mean the ability to handle github issues querying/opening inside Pharo, no?
>>
>> Thierry
>>
>> Le 23/11/2015 16:25, Skip Lentz a écrit :
>>> Hi everyone,
>>>
>>> As part of my internship I am creating bindings to the GitHub API in Pharo.
>>> As a prototype and demo, I have created a small tool last week to do
>>> some basic versioning, namely checking out a version, committing a
>>> version and showing a log of commits along with a branch tree.
>>>
>>> Here’s a screenshot: http://i.imgur.com/iMfWOvp.png
>>>
>>> The repository of the bindings and the tool is here:
>>> http://smalltalkhub.com/#!/~Balletie/GitHub
>>> To load the tool into your image, execute:
>>>
>>> (ConfigurationOfGitHub project version: #development) load: #tool
>>>
>>> Keep in mind that this is tied to GitHub, since internally it uses the
>>> API. A nice side effect of this is that everything can be done
>>> in-memory. That is, there’s no local copy on the filesystem. One does
>>> not even need git installed.
>>>
>>> That being said, feel free to take off with my prototype and make it
>>> work with e.g. the LibGit bindings in Pharo.
>>>
>>> Known bug:
>>> - When selecting a different repository from the dropdown while a
>>> version is also selected in the log, one gets a DNU. To work around this
>>> for the time being, just deselect the version before you switch
>>> repositories.
>>>
>>> Let me know what you think and feel free to ask some questions.
>>>
>>> Skip
>
>

stepharo

Re: Basic versioning of GitHub repositories from within Pharo

In reply to this post by Thierry Goubier

Hi thierry

Skip is working on a code review tool :)
and we decided after the experience of Griotte from thomas not to build
our own again but interface with existing hence
we chose git and github like tools.
We want to support code review (publish metadata) on a publish code.

Stef

Le 23/11/15 18:55, Thierry Goubier a écrit :

> Le 23/11/2015 18:40, Skip Lentz a écrit :
>>
>>> On Nov 23, 2015, at 6:32 PM, Thierry Goubier
>>> <[hidden email]> wrote:
>>>
>>> Hi Skip,
>>>
>>> this is interesting. It would mean the ability to handle github
>>> issues querying/opening inside Pharo, no?
>>
>> Yes, if that part of the API is implemented. There are still a lot of
>> extensions to be made.
>> I am planning to implement the bindings for pull requests in the near
>> future.
>
> It would be very cool to be able to envision a git + pharo combination
> where going on the github website wouldn't be necessary.
>
> Creating and commenting on an issue directly from inside Pharo instead
> of logging in fogbugz... Commenting on a commit from inside Pharo...
>
> Would you be able to retrieve the same information from github that
> GitFileTree does when it extract package version history information?
>
> Thierry
>
>

Julien Delplanque

Re: Basic versioning of GitHub repositories from within Pharo

In reply to this post by Skip Lentz-2

Awesome, I'll use it soon :)

On 23/11/15 16:25, Skip Lentz wrote:

> Hi everyone,
>
> As part of my internship I am creating bindings to the GitHub API in Pharo.
> As a prototype and demo, I have created a small tool last week to do some basic versioning, namely checking out a version, committing a version and showing a log of commits along with a branch tree.
>
> Here’s a screenshot: http://i.imgur.com/iMfWOvp.png <http://i.imgur.com/iMfWOvp.png>
>
> The repository of the bindings and the tool is here: http://smalltalkhub.com/#!/~Balletie/GitHub
> To load the tool into your image, execute:
>
> (ConfigurationOfGitHub project version: #development) load: #tool
>
> Keep in mind that this is tied to GitHub, since internally it uses the API. A nice side effect of this is that everything can be done in-memory. That is, there’s no local copy on the filesystem. One does not even need git installed.
>
> That being said, feel free to take off with my prototype and make it work with e.g. the LibGit bindings in Pharo.
>
> Known bug:
> - When selecting a different repository from the dropdown while a version is also selected in the log, one gets a DNU. To work around this for the time being, just deselect the version before you switch repositories.
>
> Let me know what you think and feel free to ask some questions.
>
> Skip

Thierry Goubier

Re: Basic versioning of GitHub repositories from within Pharo

In reply to this post by Skip Lentz-2

Hi Skip,

Trying here, especially interested to see how it handles metadata-less filetree.

- I have no authentification issues (but I had the feeling the GUI wasn't working for a little while)

- I don't get all my repositories (well, in practice allmost none, however I get all the pharo books), so I can't test that metadata-less.

- I get NeoJSON parser errors.

Nice to see that nice graph widget put to good use :) But it could have less white space on the left.

Now, how do I contribute if I want to? Where is the repository I should fork on github? I'm also a taker for a few more comments...

Thierry

2015-11-23 16:25 GMT+01:00 Skip Lentz <[hidden email]>:

Hi everyone,

As part of my internship I am creating bindings to the GitHub API in Pharo.
As a prototype and demo, I have created a small tool last week to do some basic versioning, namely checking out a version, committing a version and showing a log of commits along with a branch tree.

Here’s a screenshot: http://i.imgur.com/iMfWOvp.png

The repository of the bindings and the tool is here: http://smalltalkhub.com/#!/~Balletie/GitHub
To load the tool into your image, execute:

(ConfigurationOfGitHub project version: #development) load: #tool

Keep in mind that this is tied to GitHub, since internally it uses the API. A nice side effect of this is that everything can be done in-memory. That is, there’s no local copy on the filesystem. One does not even need git installed.

That being said, feel free to take off with my prototype and make it work with e.g. the LibGit bindings in Pharo.

Known bug:
- When selecting a different repository from the dropdown while a version is also selected in the log, one gets a DNU. To work around this for the time being, just deselect the version before you switch repositories.

Let me know what you think and feel free to ask some questions.

Skip

stepharo

Re: Basic versioning of GitHub repositories from within Pharo

In reply to this post by Skip Lentz-2

What I would like for Pharo is to avoid to be bound to a given back-end
for its versionning.
This master is a step in the right direction

http://www.hpi.uni-potsdam.de/fileadmin/hpi/source/Technische_Berichte/HPI_54.pdf

Stef

kilon.alios

Re: Basic versioning of GitHub repositories from within Pharo

And there lies the trap.

If you end up making something that works with everything, you will create something that just works with everything instead of something that works very well with one thing. Right now Git is by very far the undisputed king of version control and has completely dethroned SVN.

Ironically the things that make git integration in pharo so hard is all the thing that are there to decouple it from git , like filetree metadata , or continuous support for mcz and monticello.

In the end you cant have your cake and eat it too. We will have to choose between very good integration with git or average / mediocre integration with multiple backends.

On Sun, Nov 29, 2015 at 8:15 PM stepharo <[hidden email]> wrote:

What I would like for Pharo is to avoid to be bound to a given back-end
for its versionning.
This master is a step in the right direction

http://www.hpi.uni-potsdam.de/fileadmin/hpi/source/Technische_Berichte/HPI_54.pdf

Stef

demarey

Re: Basic versioning of GitHub repositories from within Pharo

Hi,

Le 29 nov. 2015 à 20:00, Dimitris Chloupis a écrit :

And there lies the trap.

If you end up making something that works with everything, you will create something that just works with everything instead of something that works very well with one thing.

could be. But we do not want to plug everything. Just having the door opened for the next cool VCS.

Even if is only used for git it is good to have a Pharo API that is not the GIT API. We do not want git calls mixed everywhere. We want a clean and clear API for versioning.

This API also prevents you from git API changes. You will only have one place to update and so on.

Right now Git is by very far the undisputed king of version control and has completely dethroned SVN.

King because widely used but other are also as valuable if not more.

Ironically the things that make git integration in pharo so hard is all the thing that are there to decouple it from git , like filetree metadata , or continuous support for mcz and monticello.

I do not agree on this point. There is no VCS interoperability layer in Pharo. Tools are directly manipulating Monticello.

In the end you cant have your cake and eat it too. We will have to choose between very good integration with git or average / mediocre integration with multiple backends.

again, I cannot agree. It is a bad idea to have calls to an external tool / library wide spread in the image. We need our own API. It is particularly true for git that has an API mixing daily tasks with other power admin commands (rewrite of the history with possible code loss).

I do not think we need full support of git in the image, i.e. supporting all commands and options, but only what is needed to do correct versioning from Pharo (special git commands can still be done with command-line).

On Sun, Nov 29, 2015 at 8:15 PM stepharo <[hidden email]> wrote:
What I would like for Pharo is to avoid to be bound to a given back-end
for its versionning.
This master is a step in the right direction

http://www.hpi.uni-potsdam.de/fileadmin/hpi/source/Technische_Berichte/HPI_54.pdf

Stef

smime.p7s (5K) Download Attachment

Offray

Re: Basic versioning of GitHub repositories from within Pharo

In reply to this post by kilon.alios

Hi,

On 29/11/15 14:00, Dimitris Chloupis wrote:
> And there lies the trap.
>
> If you end up making something that works with everything, you will
> create something that just works with everything instead of something
> that works very well with one thing. Right now Git is by very far the
> undisputed king of version control and has completely dethroned SVN.
>

If wisdom of the crowds were the wise path to follow, nobody would be
using Pharo/Smalltalk. The kingdom of git is because of popularity, not
from any particular technical superiority. Despite of its popularity, I
have been able to keep myself away from git/GitHub, using them only for
cloning and bug reporting, but for everything else I use fossil, a tool
that tries to be out of my way and doesn't mess a lot with my workflow
(a big difference with git) and keep metadata (bug reporting, wikis,
conversations) with me instead of using it for making me depend on
Central Server Inc. (a big difference with GitHub).

I agree with Christophe and Steph on this and is nice to have people
making this researching. A very welcomed work.

Cheers,

Offray

EuanM

Re: Basic versioning of GitHub repositories from within Pharo

In reply to this post by kilon.alios

Well, personally, I far prefer Mercurial. Which also dethroned SVN.

Mercurial has all the power of Git, while providing a more usable API.
My mind works like a Smalltalker's, not like Linus's.

It's true, some of the more abstruse functions of git require a longer
chain of user actions in Mercurial in order to achieve the same end.
But typically, the more common functions are more memorable in
Mercurial than in git, and my typical use-cases for a DSCM system are
less involved than that of the Linux core.

On 29 November 2015 at 19:00, Dimitris Chloupis <[hidden email]> wrote:

> And there lies the trap.
>
> If you end up making something that works with everything, you will create
> something that just works with everything instead of something that works
> very well with one thing. Right now Git is by very far the undisputed king
> of version control and has completely dethroned SVN.
>
> Ironically the things that make git integration in pharo so hard is all the
> thing that are there to decouple it from git , like filetree metadata , or
> continuous support for mcz and monticello.
>
> In the end you cant have your cake and eat it too. We will have to choose
> between very good integration with git or average / mediocre integration
> with multiple backends.
>
> On Sun, Nov 29, 2015 at 8:15 PM stepharo <[hidden email]> wrote:
>>
>> What I would like for Pharo is to avoid to be bound to a given back-end
>> for its versionning.
>> This master is a step in the right direction
>>
>>
>> http://www.hpi.uni-potsdam.de/fileadmin/hpi/source/Technische_Berichte/HPI_54.pdf
>>
>> Stef
>>
>>
>

EuanM

Re: Basic versioning of GitHub repositories from within Pharo

We want a system of small objects with loose-coupling, and simple webs
of message-sends. This allows us power, flexibility, maintainability
*and* the opportunity to accommodate new requirements.

We always need to bear this in mind. Build for the inevitable changes
of environment and changes of requirement. You Ain't Gonna Need it
(YAGNI) is often true, but well-factored code is *always* good.

On 30 November 2015 at 17:02, EuanM <[hidden email]> wrote:

> Well, personally, I far prefer Mercurial. Which also dethroned SVN.
>
> Mercurial has all the power of Git, while providing a more usable API.
> My mind works like a Smalltalker's, not like Linus's.
>
> It's true, some of the more abstruse functions of git require a longer
> chain of user actions in Mercurial in order to achieve the same end.
> But typically, the more common functions are more memorable in
> Mercurial than in git, and my typical use-cases for a DSCM system are
> less involved than that of the Linux core.
>
> On 29 November 2015 at 19:00, Dimitris Chloupis <[hidden email]> wrote:
>> And there lies the trap.
>>
>> If you end up making something that works with everything, you will create
>> something that just works with everything instead of something that works
>> very well with one thing. Right now Git is by very far the undisputed king
>> of version control and has completely dethroned SVN.
>>
>> Ironically the things that make git integration in pharo so hard is all the
>> thing that are there to decouple it from git , like filetree metadata , or
>> continuous support for mcz and monticello.
>>
>> In the end you cant have your cake and eat it too. We will have to choose
>> between very good integration with git or average / mediocre integration
>> with multiple backends.
>>
>> On Sun, Nov 29, 2015 at 8:15 PM stepharo <[hidden email]> wrote:
>>>
>>> What I would like for Pharo is to avoid to be bound to a given back-end
>>> for its versionning.
>>> This master is a step in the right direction
>>>
>>>
>>> http://www.hpi.uni-potsdam.de/fileadmin/hpi/source/Technische_Berichte/HPI_54.pdf
>>>
>>> Stef
>>>
>>>
>>

kilon.alios

Re: Basic versioning of GitHub repositories from within Pharo

let me give you the thinking of someone that comes from another language

"What is this ? Ah Pharo

What features it has ? Ah thats some nice cool features, live coding looks sweet

How about version control ? Hmm no its too weak and their website for hosting repos is problematic in many cases

Maybe I could use git which I am familiar with ? Ah yes.... but wait there are many limitations

How about big repos that git excels at ? Nope it crashes Pharo VM.

Hmm.. ok Pharo looks like nice toy .... will go back to my language of choice and use it from time to time for foolish experiments. Notice one thing here, the user never cares about the flexibility of the API. Bottom line its not about git, or mecurial, or svn. Its about giving something that the user can use and right now when it comes to version control Pharo is light years behind.

Things get popular not because they have nicely designed APIs, they become popular and rightly so and wisely so because they work. Simple as that.

I hope at last version control is addressed seriously because its a real pity Pharo not to be able to use git when there are toy languages like brainfuck that interface better with it or other version control system.

I am not against the idea of an abstraction API for Version Control , great idea do that, but first let get things working properly so we dont get people scratching their heads on git merges or people struggling to find out why sthub is down once again.

On Mon, Nov 30, 2015 at 7:07 PM EuanM <[hidden email]> wrote:

We want a system of small objects with loose-coupling, and simple webs
of message-sends. This allows us power, flexibility, maintainability
*and* the opportunity to accommodate new requirements.

We always need to bear this in mind. Build for the inevitable changes
of environment and changes of requirement. You Ain't Gonna Need it
(YAGNI) is often true, but well-factored code is *always* good.

On 30 November 2015 at 17:02, EuanM <[hidden email]> wrote:
> Well, personally, I far prefer Mercurial. Which also dethroned SVN.
>
> Mercurial has all the power of Git, while providing a more usable API.
> My mind works like a Smalltalker's, not like Linus's.
>
> It's true, some of the more abstruse functions of git require a longer
> chain of user actions in Mercurial in order to achieve the same end.
> But typically, the more common functions are more memorable in
> Mercurial than in git, and my typical use-cases for a DSCM system are
> less involved than that of the Linux core.
>
> On 29 November 2015 at 19:00, Dimitris Chloupis <[hidden email]> wrote:
>> And there lies the trap.
>>
>> If you end up making something that works with everything, you will create
>> something that just works with everything instead of something that works
>> very well with one thing. Right now Git is by very far the undisputed king
>> of version control and has completely dethroned SVN.
>>
>> Ironically the things that make git integration in pharo so hard is all the
>> thing that are there to decouple it from git , like filetree metadata , or
>> continuous support for mcz and monticello.
>>
>> In the end you cant have your cake and eat it too. We will have to choose
>> between very good integration with git or average / mediocre integration
>> with multiple backends.
>>
>> On Sun, Nov 29, 2015 at 8:15 PM stepharo <[hidden email]> wrote:
>>>
>>> What I would like for Pharo is to avoid to be bound to a given back-end
>>> for its versionning.
>>> This master is a step in the right direction
>>>
>>>
>>> http://www.hpi.uni-potsdam.de/fileadmin/hpi/source/Technische_Berichte/HPI_54.pdf
>>>
>>> Stef
>>>
>>>
>>

kilon.alios

Re: Basic versioning of GitHub repositories from within Pharo

just for the usual note, thats my personal opinion and in no way try to discourage people just offer a different perspective to this. I always welcome any effort to improve pharo in any way, even for ways I dont care about so much. So keep up the amazing work you guys are doing with Pharo.

On Mon, Nov 30, 2015 at 7:34 PM Dimitris Chloupis <[hidden email]> wrote:

let me give you the thinking of someone that comes from another language

"What is this ? Ah Pharo
What features it has ? Ah thats some nice cool features, live coding looks sweet
How about version control ? Hmm no its too weak and their website for hosting repos is problematic in many cases
Maybe I could use git which I am familiar with ? Ah yes.... but wait there are many limitations
How about big repos that git excels at ? Nope it crashes Pharo VM.

Hmm.. ok Pharo looks like nice toy .... will go back to my language of choice and use it from time to time for foolish experiments. Notice one thing here, the user never cares about the flexibility of the API. Bottom line its not about git, or mecurial, or svn. Its about giving something that the user can use and right now when it comes to version control Pharo is light years behind.

Things get popular not because they have nicely designed APIs, they become popular and rightly so and wisely so because they work. Simple as that.

I hope at last version control is addressed seriously because its a real pity Pharo not to be able to use git when there are toy languages like brainfuck that interface better with it or other version control system.

I am not against the idea of an abstraction API for Version Control , great idea do that, but first let get things working properly so we dont get people scratching their heads on git merges or people struggling to find out why sthub is down once again.

On Mon, Nov 30, 2015 at 7:07 PM EuanM <[hidden email]> wrote:
We want a system of small objects with loose-coupling, and simple webs
of message-sends. This allows us power, flexibility, maintainability
*and* the opportunity to accommodate new requirements.

We always need to bear this in mind. Build for the inevitable changes
of environment and changes of requirement. You Ain't Gonna Need it
(YAGNI) is often true, but well-factored code is *always* good.

On 30 November 2015 at 17:02, EuanM <[hidden email]> wrote:
> Well, personally, I far prefer Mercurial. Which also dethroned SVN.
>
> Mercurial has all the power of Git, while providing a more usable API.
> My mind works like a Smalltalker's, not like Linus's.
>
> It's true, some of the more abstruse functions of git require a longer
> chain of user actions in Mercurial in order to achieve the same end.
> But typically, the more common functions are more memorable in
> Mercurial than in git, and my typical use-cases for a DSCM system are
> less involved than that of the Linux core.
>
> On 29 November 2015 at 19:00, Dimitris Chloupis <[hidden email]> wrote:
>> And there lies the trap.
>>
>> If you end up making something that works with everything, you will create
>> something that just works with everything instead of something that works
>> very well with one thing. Right now Git is by very far the undisputed king
>> of version control and has completely dethroned SVN.
>>
>> Ironically the things that make git integration in pharo so hard is all the
>> thing that are there to decouple it from git , like filetree metadata , or
>> continuous support for mcz and monticello.
>>
>> In the end you cant have your cake and eat it too. We will have to choose
>> between very good integration with git or average / mediocre integration
>> with multiple backends.
>>
>> On Sun, Nov 29, 2015 at 8:15 PM stepharo <[hidden email]> wrote:
>>>
>>> What I would like for Pharo is to avoid to be bound to a given back-end
>>> for its versionning.
>>> This master is a step in the right direction
>>>
>>>
>>> http://www.hpi.uni-potsdam.de/fileadmin/hpi/source/Technische_Berichte/HPI_54.pdf
>>>
>>> Stef
>>>
>>>
>>

Offray

Re: Basic versioning of GitHub repositories from within Pharo

In reply to this post by kilon.alios

Hi,

On 30/11/15 12:34, Dimitris Chloupis wrote:

> let me give you the thinking of someone that comes from another language
>
> "What is this ? Ah Pharo
> What features it has ? Ah thats some nice cool features, live coding
> looks sweet
> How about version control ? Hmm no its too weak and their website for
> hosting repos is problematic in many cases
> Maybe I could use git which I am familiar with ? Ah yes.... but wait
> there are many limitations
> How about big repos that git excels at ? Nope it crashes Pharo VM.
>

Maybe I could use X because I'm familiar with is the proper argument if
is not taken too far. Interface with the external world without a
forceful coupling to X1 should be the approach.

>
> Things get popular not because they have nicely designed APIs, they
> become popular and rightly so and wisely so because they work. Simple
> as that.
>

Considering that several things work in a particular domain (data bases,
DVCS, etc) that seems reductive.

> I hope at last version control is addressed seriously because its a
> real pity Pharo not to be able to use git when there are toy languages
> like brainfuck that interface better with it or other version control
> system.
>
> I am not against the idea of an abstraction API for Version Control ,
> great idea do that, but first let get things working properly so we
> dont get people scratching their heads on git merges or people
> struggling to find out why sthub is down once again.
>

With that approach we should get Oracle intimately coupled with Pharo
before having Garage or any DB related advance. In fact I'm using
NBSQLite for *my* own purposes and needs, but I don't plan it to be the
default in the language/environment. May be because pharo goes beyond
the language and has and integrated experience, we think in how we can
make it smoother for other people, like frameworks which integrate git
or sqlite by default, but a proper balance to not enforce my particular
tastes over others should be kept also. So I would advice tightly
integration for "apps" (Scratch, grafoscopio) and loose one for
frameworks/environments (Aida, Seaside, and pharo itself).

Cheers,

Offray

stepharo

Re: Basic versioning of GitHub repositories from within Pharo

In reply to this post by demarey

+ 1

It is about abstraction.

Le 30/11/15 10:06, Christophe Demarey a écrit :

Hi,

Le 29 nov. 2015 à 20:00, Dimitris Chloupis a écrit :

And there lies the trap.

If you end up making something that works with everything, you will create something that just works with everything instead of something that works very well with one thing.

could be. But we do not want to plug everything. Just having the door opened for the next cool VCS.

Even if is only used for git it is good to have a Pharo API that is not the GIT API. We do not want git calls mixed everywhere. We want a clean and clear API for versioning.

This API also prevents you from git API changes. You will only have one place to update and so on.

Right now Git is by very far the undisputed king of version control and has completely dethroned SVN.

King because widely used but other are also as valuable if not more.

Ironically the things that make git integration in pharo so hard is all the thing that are there to decouple it from git , like filetree metadata , or continuous support for mcz and monticello.

I do not agree on this point. There is no VCS interoperability layer in Pharo. Tools are directly manipulating Monticello.

In the end you cant have your cake and eat it too. We will have to choose between very good integration with git or average / mediocre integration with multiple backends.

again, I cannot agree. It is a bad idea to have calls to an external tool / library wide spread in the image. We need our own API. It is particularly true for git that has an API mixing daily tasks with other power admin commands (rewrite of the history with possible code loss).

I do not think we need full support of git in the image, i.e. supporting all commands and options, but only what is needed to do correct versioning from Pharo (special git commands can still be done with command-line).

On Sun, Nov 29, 2015 at 8:15 PM stepharo <[hidden email]> wrote:

What I would like for Pharo is to avoid to be bound to a given back-end
for its versionning.
This master is a step in the right direction

http://www.hpi.uni-potsdam.de/fileadmin/hpi/source/Technische_Berichte/HPI_54.pdf

Stef

stepharo

Re: Basic versioning of GitHub repositories from within Pharo

In reply to this post by kilon.alios

Why could not I publish code to a fossil repo using the same
"conceptual" commands than git?
Same question for Mercurial.

So we are talking about abstraction over low-level APIs.

Basic engineering practices. This is what we are doing with Athens (do
not expose Cairo to the complete system)
OSwindow (do not expose SDL to the complete system).

Pharo is not done (but people could do that but when I see that peopel
were forced to reverse engineer the git protocol)
to script git at least this is not our objectives.

Our objectives of the mail I sent and the answers of christophe are
- make sure that we can version our code
- without be tight to a back-end

Stef