environmental challenges

[robert]

Good evening,

I have run into a number of problems, leaving me stuck, perhaps
obligated to figure it out and fix it myself. I have always admired that
principle of integrity to really take the few with the tenacity and
bring them into the community. It leaves a lot of other people to find
easier answers.

I will be using newspeak for my project, biz plan attached. I've shared
it here before, I believe, but this one has my mark, a lock, version
3.2.3 and it's size is 451 kb and its MIT so use it: a race. I've
specified the avatar and left open the RPWN to emrge the structs out.
Types of permits, visas, contracts are all open for now.

I would like a robust, helpful community, like Squeak's was back in the
day, and is now and will to come. I pray it be.

I was in last years Dec image, working my NIO and had loaded libSE &
crypto. Working with repositories I ran out of memory. Tried to load the
.ns3 I had been dumping and it blew up. I'll dig that way after I finish
this email.

I have been unable to learn how to import nor have meaningful
conversation about it. I don't see examples nor does the tutorial
contain them from my reading of it. There is a shadow on the land. Be
healed.

I tried to jump up to the latest from your download page and couldn't
load the securesockets and they 're stale I guess. It exceptioned. I'll
dig that way as well.

I am unfamiliar with your policies and procedures, as I am brand
spanking new as a member of this community. I have had super
interactions with all of you, thank you. Is there a document outlining
the procedures or the organization workflow?

Having the main download give a bleeding edge image is tough for
newcomers. Or perhaps it is securesockets with the problem. Let me go
try it again.

I was led to speak about this in hopes that more folks will come be a
part of this community, you know. Let's see it grow.

best
Robert

[robert]

Good morning,

I'll say I'm sorry to everyone for letting my frustration bleed onto the
list. I tried to speak with candor; honesty being essential, I've always
felt. I apologize to you.

I do still have some difficulty but I am hesitant to bring to the list.
I can't make heads nor tail out of the specification and just need
orientation to name resolution. Currently wondering if I can do an outer
send to the category. I see how a category is the only way to define
variableByteSubclasses.

Another issue I foresee is how to have platform neutral facade code with
platform specific on the inside, for NIO.


Best Regards,
Robert

[Gilad Bracha]

Robert,

I'll try and rephrase your question as I think I may see what the specific issue is.  

How does one access the IDE namespace, in order to feed a module appropriate arguments?

This has changed over time. I'll describe the state of affairs in bleeding edge. It is significantly different than in older versions, as categories and packages don't play as much of a role anymore. It will evolve further as the namespace facilities in the IDE should get more elaborate over time.

There are two parts to this:

(1) Defining applications. This is covered in the tutorial (see the discussion of deployment and BraveNewWorldExplorerApp).  The key thing is that the app has a factory #packageUsing: which takes a manifest object as an argument. The IDE will automatically provide a manifest object when you deploy. The API of the manifest object is essentially the set of names in the ide's root namespace.  The notion of application and its deployment is pretty stable. The actual namespace is less stable.

(2) Accessing the namespace during interactive development. The ide's root namespace is automatically in scope inside a workspace. This is achieved by having the workspace class #doesNotUnderstand: method delegate to the ide root namespace.  If you actually want to see what's there, you can evaluate 

ide namespacing Root

in a workspace and inspect it.  It includes not just classes, but other resources one might need. 

If you want to see the manifest object that gets passed to applications when they are deployed evaluate

ide namespacing manifestForSqueak

You can instantiate your module in a workspace, passing the requisite arguments to its factory. Or you can define an application and run it from the IDE. In the case of a library, the former is more natural.

You could also try inspecting the namespace presenter, or look at the Workspace class to see how they are set up. The entire system is inspectable. The insect presenter menu option is available on most of the UI and lead you to the actual implementations of things. 

If this wasn't the question then I'm sorry.

[robert]

I have a few concerns:

1) why are modules dependent on IDE? What about headless?
2) why aren't categories in the namespace lookup path?
3) new NSers need a cheat sheet without a lot of verbage or formalisms. Just groupings of use (assignments, condionals, looping, namespace access, initializations, ...) with examples. The spec doesn't do it and the tutorial helps a lot but is very verbose - it's a tutorial. AS cheat sheet is different. A poor-mans reference, if you would.
4) clearer vision, mission & strategy. IDE Apps? Mobile Apps? Headless Apps? broad user base? stovepiped efforts as each chase their own passions?
5) define & assign clearer roles to support 4

That would be my take.

[robert]

Whatever is defined for 4 and 5, 6) should be the following, as I wrote you privately:

6) organize around the user of the language to develop them into stars.
7) be welcoming and inviting to hackers.

For example, did you know that the Address Family of a socket creation can be Bluetooth?

[Gilad Bracha]

On Thu, Sep 24, 2015 at 1:12 PM robert <robert.w...@gmail.com> wrote:

Whatever is defined for 4 and 5, 6) should be the following, as I wrote you privately:

6) organize around the user of the language to develop them into stars.
7) be welcoming and inviting to hackers.

Thank you for your input.  

For example, did you know that the Address Family of a socket creation can be Bluetooth?

 And this is relevant to this discussion in some way I suppose. 

 

On 9/24/2015 4:03 PM, robert wrote:

I have a few concerns:

1) why are modules dependent on IDE?

They aren't.  Modules have no external dependencies. Applications are analogous to build scripts, and they depend on some environment no matter what language you use.  They are like makefiles.

 

What about headless?

A headless build environment would be a script that provides an application with a manifest object. It could create that based on directory structure. environment variables, URIs or whatever. All of which is orthogonal to whether the actual application is headless.

 

2) why aren't categories in the namespace lookup path?

They used to be, but we don't want a dependency on that.

3) new NSers need a cheat sheet without a lot of verbage or formalisms. Just groupings of use (assignments, condionals, looping, namespace access, initializations, ...) with examples. The spec doesn't do it and the tutorial helps a lot but is very verbose - it's a tutorial. AS cheat sheet is different. A poor-mans reference, if you would.

I probably should write one.  It would take less time than answering individual questions repeatedly.

4) clearer vision, mission & strategy. IDE Apps? Mobile Apps? Headless Apps? broad user base? stovepiped efforts as each chase their own passions?

That may be true.  

 

5) define & assign clearer roles to support 4

We might do that, when and if we have people who we are convinced competent and trustworthy and who have the time to engage in that.

[robert]

I was asked and those were my thoughts about it, from an outsiders perspective. What do other folks think is the way forward? The time is ripe.

Best
Robert

[robert]

Look at how you responded to what I wrote, another chance. 6 + 7 + bluetooth = awesome and you don't connect it. You meed to check your parachute's color and be honest with yourself; are you playing to your strengths? What's your general opinion of that.  I'm not sure I see it the same.

On 9/24/2015 4:29 PM, Gilad Bracha wrote: