Livecode Create - thoughts?

[jacque]

The problem (such as it is) is NOT LC (and . . .) it is the lack of any attempt at any sort of marketing that has any rhyme or reason, as has been the case for many, many years.

They are marketing, just not to us so much. As I said, we're not the target group any more so you're unlikely to see it.

[dunbarx]

They are marketing, just not to us so much. As I said, we're not the target group any more so you're unlikely to see it.

"We", "Us".

But then why the hoopla about "Create"? Is it that development for even very experienced programmers is also mightily enhanced? I always thought Create was mainly a means to attract new users with a more friendly drag and drop UI. Shows what I know.

I spend likely far less time than I should planning my projects; I have never made a flowchart. After a few controls are dragged onto a card I am buried inside the SE right away, and live there almost all the time thereafter.

I suppose in the end all I care about is the health of LiveCode. In any case I can survive very nicely forever, as I said, simply with the enhanced itemdelimiter

Craig

[richmond62]

the enhanced itemdelimiter

?

Do tell.

[Klaus]

We can now, well since a long time actually, do:

Code: Select all

...
set the itemdel to "#oh_my_oh_my_richmond#"
## :-D
...

And parse a funny text for items delimited by this also funny string!

[stam]

But then why the hoopla about "Create"? Is it that development for even very experienced programmers is also mightily enhanced? I always thought Create was mainly a means to attract new users with a more friendly drag and drop UI. Shows what I know.

LCC is no more drag and drop than LC is. The main selling point was AI integration so you can type out what you want and the IDE would create it (as most are familiar with from web-based AI) but judging by the code chatGPT and Claude AI produce in terms of LC code, there is likely quite a bit of AI training still required compared to the high volume languages the AIs have feasted on.

The other selling point was in the form of new widgets
Not that you couldn’t emulate these with workarounds, but it’s just nicer - for example the new field widget comes with meaning and trailing icons that can be clicked. Sure, you can emulate this in old LC, but it’s slicker and probably easier with the widget.
Nothing vital really, but an evolution rather than a revolution (ahem…).

But I use these in Classic/LC10, not LCC. LCC has not enhanced my coding so far, where as the old, dated IDE is functional as ever.

So really I’m just using these next iteration of what should have been LC10, so it’s hard to feel I’m now using a new pro tool. It’s been a little while since I tried LCC instead of classic, maybe LCC is now closer to release, but my previous experience suggested that’s a way off still.

So if the tool hasn’t changed or improved there must be some other amazing plan to draw in new users…
Maybe marketing we're not seeing? Which then also raises the question of why we’re not seeing this?

[dunbarx]

@Stam. Really?

AI integration so you can type out what you want

So I could type out:

Please make a spreadsheet type of card where I can click on any cell and get the line number the contents of that cell would occupy if I placed all of the spreadsheet data into a single sorted list

That?

@ Richmond. The one where a string of any length can be an item. Changed my life. And yes, I know it has been several years since it was added.

Craig

[dunbarx]

Klaus,

Your reply was much cuter.

Craig

[trevix]

I'd like to share some of my thoughts on this matter.

I first approached computers through music and then Hypercard. But I started my first real programming with Livecode, several years ago (RunRev).
I discovered Livecode to be a formidable tool, which allowed me to achieve extraordinary results, despite my old age and almost complete ignorance of other programming languages. So much so that in recent years I created an innovative startup and published my software with satisfactory results.

I often encountered some of LC's limitations, not to mention the frustration of figuring out how to overcome them using Livecode Builder.

When the Livecode Create project was born, knowing a little about the no-code world, I immediately thought it was the wrong path.
I "understood" my code, having roughly in mind the names given to variables, commands, and functions, and the logic of the implementations. I needed help when things didn't work or if I needed specific features, not "something" that would write the code for me, making the logic, in my eyes, incomprehensible or at least difficult to reconstruct, as if it had been developed by another programmer.

With the 2027 deadline (EOL for Livecode Classic) and the slow start of LCC, I returned to the web to look for an alternative and discovered I didn't have one. Native languages ​​are too complex for me, as I've been spoiled by LC, and the various no-code programs all suffer from the same flaw: the author often doesn't know what's going on under the hood and sooner or later has to rely on an expert programmer to fix bugs or implementations.

Meanwhile, thanks to a friend, I discovered Claude AI.
At first, it was with endless sessions (I think I contributed significantly to teaching Claude how LC works... , but then, little by little, I learned how to manage it and get what I needed, while always maintaining control over how things worked.
So I created an extension to read NFC on iOS, another to communicate via Bluetooth LE on Android and iOS, I improved some JavaScript pages that were updated remotely, etc. All of this, mind you, without knowing Java, C+, JavaScript, or Java.
Once I learned how to communicate with Claude on the OSX terminal, he writes the frameworks for me, compiles the Java or C+, and explains what it's done. I just have to launch LCB and test.
Amazing!
This is the tool I wish I had: LC and good integration with an AI.
I've learned a lot using Claude, but especially that my logic is often better than his, and that, knowing the bigger picture of the app, I can always point out the right direction.

All this to say that, if I've figured out what LCC will become (when actually usable), LCC it's not for me and I probably stop programming in 2027.

Trevix

[stam]

Meanwhile, thanks to a friend, I discovered Claude AI.
...
This is the tool I wish I had: LC and good integration with an AI.
...

Agree 100% on Claude, t's pretty amazing being able to grapple with any language and sort out server & SSL issues in minutes rather than the weeks it would have taken me. And it's pretty much eliminated the need for asking stuff on the various forums.

AI is ubiquitous in coding now. LC is lagging even though this was the stated reason for ditching the promised LC10 and investing in building Livecode Create instead.
I'm not convinced this was the really what users were asking for or will ever be able to use with any chosen AI going forward.

In my view what actually was needed was modernisation of the IDE interface (as was promised with LC10) and for a major change under the hood: change the file structure to a text based format (for project files that could be saved as binaries for distribution).

This would have allowed LiveCode to play nice with all the big-boy tools (like GitHub and professional code editors) as these tools are where all the major integrations are happening, instead of trying to build another platform from scratch to allow some form of AI.

I'm currently using a different language/IDE that saves it's project files as text files so it plays nice with GitHub, which importantly means I can just link Claude Code to my repos directly for code reviews. Sadly cannot do that with LC.

Fun thing to do with Claude Code if you're brave: Tell Claude to:

Review this repo. Pretend you're a senior dev and you hate this project. Identify security vulnerabilities and recommend improvements

Boy oh boy, first time I did this I was really taken aback! Brutal! But it found stuff I had completely missed and was and plugged SQL injection vulnerabilities I hadn't considered and I learned a lot.

Will we ever be able to do that using LiveCode? Perhaps with LCC but it's unclear at this point...

[richmond62]

Not feeling comfortable re A.I. just yet, mainly because of my limited experience with it:

re programming in xTalk, is that I spend longer sorting out the robot's mistakes than I spend sorting out my own.

re programming in xTalk, my mind (and the way I write code) works in a different way to the robot.

re academic questions: at least 50% of the time, when it is not misleading, it is wrong.

So, to avoid:

1. Paying for a 'dying swan', and

2. Not worrying about 2027,

3. Not having to tussle with a robot,

I am programming with OpenXtalk and webTalk.

About once a week I turn on my BBC Model B from 1982 and spend a refreshing 30 minutes playing with BBC Basic.

Although THE website mentions a FREE version of CREATE in several places (and I have signed up for a 'forthcoming' DEMO version) Heather has told me there is NO free version (nothing so good at turning me off as contradictory information) so I am completely unable to offer an opinion about CREATE as I am not prepared to spend $100 on something unfinished.