Comments on 0.94

[tperry2x]

You folks have done wonders on the Linux side, so I'm asking, is it possible to stand up a compiler for 963 on Linux, right? Or does it have to be compiled (for now) on another OS?

It depends on whether you are wanting to compile the IDE or the engine.
I know you specified the IDE above in your original post, but as is becoming more evident, the bugs we are encountering lately are rooted in the engine.

So, if you are compiling the IDE, then as Richmond rightly says - I think you can do that on up-to-date hardware and OS.
If you are compiling the engine, I think that's a hideous task of dependency hell involving lots of modifications to GYP.

How are you with C++, Java and Python by the way? I ask as I'm hoping you can get GYP working (as I'm quite sure that's used in-turn to build the engine).

https://www.openxtalk.org/forum/viewtopic.php?f=9&t=600

[foxtrot47]

You folks have done wonders on the Linux side, so I'm asking, is it possible to stand up a compiler for 963 on Linux, right? Or does it have to be compiled (for now) on another OS?

It depends on whether you are wanting to compile the IDE or the engine.
I know you specified the IDE above in your original post, but as is becoming more evident, the bugs we are encountering lately are rooted in the engine.

So, if you are compiling the IDE, then as Richmond rightly says - I think you can do that on up-to-date hardware and OS.
If you are compiling the engine, I think that's a hideous task of dependency hell involving lots of modifications to GYP.

How are you with C++, Java and Python by the way? I ask as I'm hoping you can get GYP working (as I'm quite sure that's used in-turn to build the engine).

https://www.openxtalk.org/forum/viewtopic.php?f=9&t=600

I wanted to compile the engine. I know next to nothing about C++, Java and Python, but I've had some "wins" in the past with an enterprise monolith or two, so I figured it was worth a shot.

I'm pretty sure GYP is what stopped me over the Summer. I remember getting it partially/fully installed and the make files being generated, but I kept getting an error that was too obscure for me to troubleshoot with online references and AI.

However, I'm still down to try again. I really like the idea of getting the engine compiler working in a VirtualBox VM, like an easy to deploy development environment. I don't know much about the process to know if that's even a good idea. Still, I enjoy punishing myself with cryptic errors and endless troubleshooting. Sometimes things work by accident. *fingers crossed*

I'll let you know if I come up with anything, but I'll drop it in here instead: https://www.openxtalk.org/forum/viewtopic.php?f=9&t=600

[richmond62]

I wanted to compile the engine.

Brave person: more strength to your elbow.

What I do not understand is 2 things:

1. It was claimed, many long years ago, that Runtime Revolution was built in Runtime Revolution . . .

2. How much (if not all) of the engine is written in C++?

If #1 is true, when did that stop being true?

[richmond62]

Am I bonkers, or am I bonkers . . .

https://www.dcpehvpm.org/E-Content/BCA/ ... %20Cpp.pdf

And how the Flying Fekk am I going to leverage that to poke around in . . .?

And where IS the '?' I can poke around in?

And, if I do more than poke around: I mean get 'clever' and actually effect a change in the '?' how on earth do I get that to someone who can 'do; a build?

61 years and counting, and naive as a cabbage.
-

Screenshot 2023-11-22 at 20.37.47.png (803.35 KiB) Viewed 1965 times

-
I mean, why would you use A.I. for anything than generate silly pictures of blue Hindu gods on motorbikes?

Brings a whole new meaning to "Push the pedal to the metal." nicht?

[tperry2x]

It was claimed, many long years ago, that Runtime Revolution was built in Runtime Revolution . . .

It's mostly true. The IDE is pretty much written inside LCC, the engine, notsomuch.

How much (if not all) of the engine is written in C++?

Lots of it.... all of it?

If #1 is true, when did [2] stop being true?

I'm not sure that the engine was ever written outside of C++ from what I can tell. I may be wrong if you look back at classic mac days.

[tperry2x]

I mean, why would you use A.I. for anything than generate silly pictures of blue Hindu gods on motorbikes?

AI seems to be the buzzword as of late, but this generated picture shows a lot wrong with the algorithm. Any human drawing that would realise that sitting side-saddle like that would result in 3rd degree burns from the engine with those robes on, the saddle bags are positioned on the exhaust, there's an irregular wheel spoke pattern which would be bumpy and dangerous to ride it, and it's got no brakes. But other than that, the AI did what an AI does and leaves out the small details.

AI seems to be an attempt to take humans out of the loop, under the guise that it'll save money and be this miraculous panacea to fix everything.

[richmond62]

That picture says buckets about the state of A.I. at the moment: anyone trusting it to do anything is going to get their robes tied up in their spokes.

[tperry2x]

We go back in a loop to https://www.openxtalk.org/forum/viewtop ... 2817#p2817
I think 'human input' of code is far far better, and if you want to ask someone why they've made a particular decision, they can show their workings-out. An AI just makes decisions based on a dataset and it's a mystery as to why.

[richmond62]

Anyway . . .

Richmond is waiting on a 0.94 for Mac so he can be thoroughly obnoxious over the weekend . . .

My 'intelligence' (such as it is) is 'organic' . . . i.e.remarkably similar to that pile of cow dung you can see in the field over the way from your sittingroom.

[tperry2x]

Richmond is waiting on a 0.94 for Mac so he can be thoroughly obnoxious over the weekend . . .

This may have been lost in communication between various posts, but you know it's already there, right?
https://www.openxtalk.org/forum/viewtop ... 4646#p4646

I'll brace myself for feedback, but many of my responses might be "it's in the engine".

[foxtrot47]

I may have forgotten to mention these in 0.94:

• Home/toolbar (first launch) opens on a monitor that is not my default monitor, and my preferred placement is not retained between sessions.
• The grid is not snapping to the number defined in Preferences > Appearance > Grid spacing. Instead, the grid seems to be fixed to 8 pixels (for me).
• Preferences > Appearance > Grid spacing: reverts to "2" when Preferences is re-opened.

Also, the bug in the Grid spacing field has been a gripe of mine for years. I am elated that it's finally receiving some much needed attention!

[tperry2x]

I may have forgotten to mention these in 0.94:

• Home/toolbar (first launch) opens on a monitor that is not my default monitor, and my preferred placement is not retained between sessions.
• The grid is not snapping to the number defined in Preferences > Appearance > Grid spacing. Instead, the grid seems to be fixed to 8 pixels (for me).
• Preferences > Appearance > Grid spacing: reverts to "2" when Preferences is re-opened.

Also, the bug in the Grid spacing field has been a gripe of mine for years. I am elated that it's finally receiving some much needed attention!

Thanks for letting me know about the grid spacing.
I know Richmond did some work on getting the values to appear correctly in the fields, but I'll revisit this.

The toolbar at the top appearing on a different monitor - is this on Linux? That's very strange, but does it appear at the same co-ordinates each time, or a seemingly random location?
Also, is it consistent as to what screen it appears on, or is the behaviour sometimes reversed with the toolbar and rest of the palettes swapping location?

[richmond62]

The thing anent the stacks opening on a monitor that is not the end-user's main monitor is odd.

All my development computers for the last 20 years have had at least 2 monitors connected, and the ONLT time I have ever had a MC/RR/LC/OXT stack opening anywhere NOT on my main monitor is when I have moved it to another monitor in a previous session. For the sake of argument, I always have the Dictionary open on my right-hand monitor so it does not overlap my 'playground'.
-

Screenshot 2023-11-24 at 9.52.49.png (184.42 KiB) Viewed 1908 times

-
Strictly for the broadminded: https://youtu.be/-z0UIpaKmPU
-
Anyway that can be sorted out with a simple script in the Home.livecodescript file to send the Menu Bar and the Tools to their 'standard' positions on the main monitor.

Always remember that ant leftwards monitors in your rig will be referred to by negative numbers in xTalk: so

Code: Select all

set the left of stack "HereWeGo" to -300

will carry the left side of your stack to somewhere on a leftward monitor, while

Code: Select all

set the left of stack "HereWeGo" to 0

will tuck your stack neatly up against the left side of your main monitor.

[tperry2x]

Think I might know why, or know what we can look at.
Line 26 to 32 of "revmenubar.livecodescript"

Code: Select all

 -- mac specific fix for placement
   if the platform contains "mac" then set the top of wd "revMenuBar" to 22
   -- linux & win specific fix for placement (going top left - should ignore menubar and panels in a good linux distro!) -- tperry
   if the platform contains "linux" or the platform contains "win" then 
      set the top of wd "revMenuBar" to 0
      set the left of  wd "revMenuBar" to 0
   end if

If your problem is on Linux, then I'd suggest that the top and left aren't being reported correctly for some reason.
I do remember I had some weirdness with linux on this, as I'd normally just set topleft, but had to set these one after eachother as otherwise it ignored it... which was weird.

[TerryL]

Well done updating Lite to .94 with many improvements! I hoped to post something positive as well but struck out on three projects.

1) Update Sqlite. The latest version of sqlite: 2023-11-01 - Version 3.44.0, https://www.sqlite.org/draft/download.html
The download has an sqlite3.js and sqlite3.wasm file but I couldn't find these anywhere in the OXT Lite or LCC 9.6.3 folders. I have a sinking feeling they're in the Engine.

2) Geometry Manager LC bug 23122: http://quality.livecode.com/
I'm not able to demo a bug. I use GM and have not had problems on Win10. Others avoid it and use on resizeStack. Both are compatible together if pass resizeStack is included at the bottom of the message handler. Give an object geometry settings and then move it to a different location or resize. Navigate to another card and back or resize the stack, and the object jumps back to it's original position or size where the GM settings were made.
The solution is to run in msg box before nav or resizing stack: revCacheGeometry. The dictionary on revCacheGeometry is clear if you know to look for it. Note, revCacheGeometry refreshes GM settings only for the current card. I understand how this would upset people who carefully reposition/resize objects to have them jump around. Maybe the bug shows on Mac or Linux.
From bug 23122 text: "...I make sure to click some other already existing object (any will do), go into its geometry manager settings, and click here and there to change and restore them. That will update the geometry properly. Then I save the stack, and the stack is stable."

3) OXT Lite Evaluation By Local High School Computer Science Class
I received no email response inviting two local high school computer science instructors to evaluate OpenXTalk Lite .93 with their students and offer suggestions for improvement. Local school district superintendents decide what computer programs and curriculum are used for grades K-12. I suspect a superintendent relies upon high school comp-sci instructors to advise. I know one high school uses Scratch. A superintendent somewhere might give OXT a try, but they are mandated to use "banding" where the coding program and techniques taught in one grade are used and added to in the next grade, up to high school grade 12. Inserting a new coding program into existing curriculum may be considered too disruptive to instructors and students.
----

[richmond62]

I received no email response inviting two local high school computer science instructors to evaluate OpenXTalk Lite .93 with their students and offer suggestions for improvement.

Well, were I a high-school programming teacher my first questions would go like this:

1. What is in this for me?

2. Why on earth should I get up to speed in this unknown language when I have spent ages getting semi-competent in Python, SCRATCH, or whatever, which is what the education authority wants me to use, and there is a wide range of textbooks available for?

Certainly, 20 years ago when I approached a Primary and a Secondary school (in St. Andrews, in Fife) the ONLY reason I got a foot in the door at the Primary school was because I knew the headmistress well, she having had my own children in her school for 3 years, and the Secondary school because the vice-principle had once worked with a second cousin of mine in a school in Dundee. And the teachers who came to my focus sessions came with their prejudices and presuppositions out in front of them like a woman pregnant with triplets: I was wasting their time, but they came along because they had been mandated to by a higher authority.

I suspect the ONLY chance you have is if any of those schools have out-of-hours programming clubs where the supervisors might be prepared to 'risk' something extra-curricular. Most school teachers, in about 35 years of experience, are a narrow bunch more concerned with getting their charges to pass exams than actually learn anything useful or innovative, whatever the latest 'blah' that the government may be promoting.

[tperry2x]

..Give an object geometry settings and then move it to a different location or resize. Navigate to another card and back or resize the stack, and the object jumps back to it's original position or size where the GM settings were made.
The solution is to run in msg box before nav or resizing stack: revCacheGeometry. The dictionary on revCacheGeometry is clear if you know to look for it. Note, revCacheGeometry refreshes GM settings only for the current card. I understand how this would upset people who carefully reposition/resize objects to have them jump around.
Maybe the bug shows on Mac or Linux.

I've seen this bug on MacOs, Windows, and Linux for as long as I can remember unfortunately. As a consequence, I never use the geometry manager. I always set it in script like so:

geom.png (25.18 KiB) Viewed 1887 times

I'm not saying that's ideal, far from it, however it's a solution that works across all 3 platforms flawlessly.

[tperry2x]

• The grid is not snapping to the number defined in Preferences > Appearance > Grid spacing. Instead, the grid seems to be fixed to 8 pixels (for me).
• Preferences > Appearance > Grid spacing: reverts to "2" when Preferences is re-opened.

Fixed, as of 25-11-23 - this will appear in the 0.95 release

[*]Home/toolbar (first launch) opens on a monitor that is not my default monitor, and my preferred placement is not retained between sessions.

I'm thinking of having a few options to "nudge" these under preferences > compatibility
edit: done (should help people with multiple monitors), this will be in the 0.95 version

screenshot.png (47.72 KiB) Viewed 1867 times