- Welcome to Interactive PowerBasic Forum.
Your resource for programming with PowerBASIC™
Recent posts
#1
AI / ChatGPT / Claude AI / LM Studio / AGI / Leveraging Multiple AI Models ...
Last post by Theo Gottwald - April 24, 2025, 02:21:24 PM**English Version with Emojis:**
**Title: How to Maximize Productivity with Google AI, OpenAI, and Grok in Your Coding Workflow 🚀💻🤖**
In recent days, I've been working on increasingly complex projects, particularly involving AI agents. For testing and development, I primarily rely on Python 🐍🖥�, as AI models excel in that environment. During this process, I've observed notable differences between various AI models, each with their own strengths and limitations. ⚖️✨
Google's latest AI model is truly impressive, boasting a massive 1MB context window 📊—unmatched for generating extensive code. It can output around 60,000 characters in one go ✍️, making it ideal for writing large final scripts. Because of this, I often start my workflow with Google AI Studio 🌐🤖, especially when I need to produce comprehensive code snippets or entire functions efficiently.
However, when it comes to troubleshooting tricky bugs 🐞🔧—those elusive issues that typically require deep dives into forums 💬, specialist sites 🌍, or even foreign-language resources 🈶—Google's AI sometimes falls short. Despite its vast knowledge base, Google's AI can be outdated in niche or complex cases 📉⚠️. That's where OpenAI's latest models, O3 and O4, shine ✨💡. Although their output length is limited and they might cut off code prematurely ✂️, their research and problem-solving capabilities are exceptional 🧠🔍. They often identify issues that Google's AI misses 🧐, making them invaluable for debugging and troubleshooting.
Grok 3 is another noteworthy model ✅. It manages longer code better 💾, produces fewer errors ❌🚫, and has more current knowledge compared to Google's AI 📚. When Google or other models reach their limits 🚧, Grok 3 can often provide critical insights 💡 and help resolve complex problems 🧩. For advanced debugging and research tasks, OpenAI's O3 and O4 models have truly revolutionized the process 🔄🚀. You can simply paste an error message or complicated problem into O3 📝, and it reliably pinpoints issues that are nearly impossible to uncover manually 🕵��♂️. It's akin to having an expert investigator at your fingertips 🔎🤝.
My current workflow involves leveraging OpenAI's models to analyze and diagnose problems 🧰🧠. Once I get the insights or solutions 💡, I paste them into Google AI Studio 🌎🤖 to generate polished, final code ✨📝. The result is code that typically runs smoothly and is free of errors 🔄✅. Additionally, you might consider combining Grok with OpenAI or switching between Google and OpenAI depending on the specific task 🔄🎯. While Grok lacks Google's enormous context window 🚫📊, it is often more up-to-date with recent developments 📅📰.
Bottom line: Combining the unique strengths of these AI models can significantly improve your coding efficiency 💪💻. Recognize each model's capabilities—whether it's Google's extensive code generation 🖥�, OpenAI's superior troubleshooting 🛠�, or Grok's up-to-date knowledge 📖—to optimize your programming workflow 🚀.
Feel free to experiment with these tools and discover the most effective combination for your projects 🧪🔍.
#AIagents 🤖 #AICoding 💻 #OpenAIO3 💡 #GoogleAI 🌐 #Grok3 🧠
---
**Versión en Español con Emojis:**
**Título: Cómo Maximizar la Productividad con Google AI, OpenAI y Grok en tu Flujo de Trabajo de Programación 🚀💻🤖**
En los últimos días, he estado trabajando en proyectos cada vez más complejos, especialmente involucrando agentes de IA 🧠🤖. Para pruebas y desarrollo, principalmente confío en Python 🐍🖥�, ya que los modelos de IA sobresalen en ese entorno. Durante este proceso, he observado diferencias notables entre varios modelos de IA, cada uno con sus propias fortalezas y limitaciones ⚖️✨.
El último modelo de IA de Google es realmente impresionante, con una ventana de contexto de 1MB 📊—inalcanzable para generar código extenso. Puede producir alrededor de 60,000 caracteres en una sola vez ✍️, lo que lo hace ideal para escribir scripts grandes o funciones completas de manera eficiente 🌐🤖. Por eso, generalmente comienzo mi flujo de trabajo con Google AI Studio 🛠�🌟, especialmente cuando necesito producir fragmentos de código completos o funciones de manera rápida.
Sin embargo, cuando se trata de solucionar bugs complicados 🐞🔧—esos problemas difíciles que suelen requerir una búsqueda profunda en foros 💬, sitios especializados 🌍, o incluso recursos en idiomas extranjeros 🈶—el IA de Google a veces no llega a la altura. A pesar de su vasta base de conocimientos, el IA de Google puede estar desactualizado en casos específicos o complejos 📉⚠️. Ahí es donde los últimos modelos de OpenAI, O3 y O4, brillan ✨💡. Aunque su longitud de salida es limitada y pueden cortar el código prematuramente ✂️, sus capacidades de investigación y resolución de problemas son excepcionales 🧠🔍. A menudo identifican problemas que el IA de Google pasa por alto 🧐, convirtiéndolos en herramientas valiosísimas para depuración y solución de problemas.
Grok 3 es otro modelo destacado ✅. Maneja mejor el código largo 💾, produce menos errores ❌🚫 y tiene conocimientos más actualizados en comparación con el IA de Google 📚. Cuando Google u otros modelos alcanzan su límite 🚧, Grok 3 puede ofrecer ideas críticas 💡 y ayudar a resolver problemas complejos 🧩. Para tareas avanzadas de depuración e investigación, los modelos O3 y O4 de OpenAI han revolucionado verdaderamente el proceso 🔄🚀. Solo necesitas pegar un mensaje de error o problema complicado en O3 📝, y con fiabilidad identifica problemas que son casi imposibles de detectar manualmente 🕵��♂️. Es como tener a un investigador experto a tu alcance 🔎🤝.
Mi flujo de trabajo actual implica utilizar los modelos de OpenAI para analizar y diagnosticar problemas 🧰🧠. Una vez que obtengo las ideas o soluciones 💡, las pego en Google AI Studio 🌎🤖 para generar código refinado y final ✨📝. El resultado es un código que generalmente funciona sin problemas y sin errores 🔄✅. Además, puedes considerar combinar Grok con OpenAI o alternar entre Google y OpenAI según la tarea específica 🔄🎯. Aunque Grok no tiene la enorme ventana de contexto de Google 🚫📊, a menudo está más actualizado con los avances recientes 📅📰.
En resumen: combinar las fortalezas únicas de estos modelos de IA puede mejorar significativamente tu eficiencia en programación 💪💻. Reconoce las capacidades de cada uno—ya sea la generación extensa de código de Google 🖥�, la mejor resolución de problemas de OpenAI 🛠�, o los conocimientos actualizados de Grok 📖—para optimizar tu flujo de trabajo de programación 🚀.
Siéntete libre de experimentar con estas herramientas y descubrir la combinación más efectiva para tus proyectos 🧪🔍.
#AgentesIA 🤖 #CodificaciónIA 💻 #OpenAIO3 💡 #GoogleAI 🌐 #Grok3 🧠
**Title: How to Maximize Productivity with Google AI, OpenAI, and Grok in Your Coding Workflow 🚀💻🤖**
In recent days, I've been working on increasingly complex projects, particularly involving AI agents. For testing and development, I primarily rely on Python 🐍🖥�, as AI models excel in that environment. During this process, I've observed notable differences between various AI models, each with their own strengths and limitations. ⚖️✨
Google's latest AI model is truly impressive, boasting a massive 1MB context window 📊—unmatched for generating extensive code. It can output around 60,000 characters in one go ✍️, making it ideal for writing large final scripts. Because of this, I often start my workflow with Google AI Studio 🌐🤖, especially when I need to produce comprehensive code snippets or entire functions efficiently.
However, when it comes to troubleshooting tricky bugs 🐞🔧—those elusive issues that typically require deep dives into forums 💬, specialist sites 🌍, or even foreign-language resources 🈶—Google's AI sometimes falls short. Despite its vast knowledge base, Google's AI can be outdated in niche or complex cases 📉⚠️. That's where OpenAI's latest models, O3 and O4, shine ✨💡. Although their output length is limited and they might cut off code prematurely ✂️, their research and problem-solving capabilities are exceptional 🧠🔍. They often identify issues that Google's AI misses 🧐, making them invaluable for debugging and troubleshooting.
Grok 3 is another noteworthy model ✅. It manages longer code better 💾, produces fewer errors ❌🚫, and has more current knowledge compared to Google's AI 📚. When Google or other models reach their limits 🚧, Grok 3 can often provide critical insights 💡 and help resolve complex problems 🧩. For advanced debugging and research tasks, OpenAI's O3 and O4 models have truly revolutionized the process 🔄🚀. You can simply paste an error message or complicated problem into O3 📝, and it reliably pinpoints issues that are nearly impossible to uncover manually 🕵��♂️. It's akin to having an expert investigator at your fingertips 🔎🤝.
My current workflow involves leveraging OpenAI's models to analyze and diagnose problems 🧰🧠. Once I get the insights or solutions 💡, I paste them into Google AI Studio 🌎🤖 to generate polished, final code ✨📝. The result is code that typically runs smoothly and is free of errors 🔄✅. Additionally, you might consider combining Grok with OpenAI or switching between Google and OpenAI depending on the specific task 🔄🎯. While Grok lacks Google's enormous context window 🚫📊, it is often more up-to-date with recent developments 📅📰.
Bottom line: Combining the unique strengths of these AI models can significantly improve your coding efficiency 💪💻. Recognize each model's capabilities—whether it's Google's extensive code generation 🖥�, OpenAI's superior troubleshooting 🛠�, or Grok's up-to-date knowledge 📖—to optimize your programming workflow 🚀.
Feel free to experiment with these tools and discover the most effective combination for your projects 🧪🔍.
#AIagents 🤖 #AICoding 💻 #OpenAIO3 💡 #GoogleAI 🌐 #Grok3 🧠
---
**Versión en Español con Emojis:**
**Título: Cómo Maximizar la Productividad con Google AI, OpenAI y Grok en tu Flujo de Trabajo de Programación 🚀💻🤖**
En los últimos días, he estado trabajando en proyectos cada vez más complejos, especialmente involucrando agentes de IA 🧠🤖. Para pruebas y desarrollo, principalmente confío en Python 🐍🖥�, ya que los modelos de IA sobresalen en ese entorno. Durante este proceso, he observado diferencias notables entre varios modelos de IA, cada uno con sus propias fortalezas y limitaciones ⚖️✨.
El último modelo de IA de Google es realmente impresionante, con una ventana de contexto de 1MB 📊—inalcanzable para generar código extenso. Puede producir alrededor de 60,000 caracteres en una sola vez ✍️, lo que lo hace ideal para escribir scripts grandes o funciones completas de manera eficiente 🌐🤖. Por eso, generalmente comienzo mi flujo de trabajo con Google AI Studio 🛠�🌟, especialmente cuando necesito producir fragmentos de código completos o funciones de manera rápida.
Sin embargo, cuando se trata de solucionar bugs complicados 🐞🔧—esos problemas difíciles que suelen requerir una búsqueda profunda en foros 💬, sitios especializados 🌍, o incluso recursos en idiomas extranjeros 🈶—el IA de Google a veces no llega a la altura. A pesar de su vasta base de conocimientos, el IA de Google puede estar desactualizado en casos específicos o complejos 📉⚠️. Ahí es donde los últimos modelos de OpenAI, O3 y O4, brillan ✨💡. Aunque su longitud de salida es limitada y pueden cortar el código prematuramente ✂️, sus capacidades de investigación y resolución de problemas son excepcionales 🧠🔍. A menudo identifican problemas que el IA de Google pasa por alto 🧐, convirtiéndolos en herramientas valiosísimas para depuración y solución de problemas.
Grok 3 es otro modelo destacado ✅. Maneja mejor el código largo 💾, produce menos errores ❌🚫 y tiene conocimientos más actualizados en comparación con el IA de Google 📚. Cuando Google u otros modelos alcanzan su límite 🚧, Grok 3 puede ofrecer ideas críticas 💡 y ayudar a resolver problemas complejos 🧩. Para tareas avanzadas de depuración e investigación, los modelos O3 y O4 de OpenAI han revolucionado verdaderamente el proceso 🔄🚀. Solo necesitas pegar un mensaje de error o problema complicado en O3 📝, y con fiabilidad identifica problemas que son casi imposibles de detectar manualmente 🕵��♂️. Es como tener a un investigador experto a tu alcance 🔎🤝.
Mi flujo de trabajo actual implica utilizar los modelos de OpenAI para analizar y diagnosticar problemas 🧰🧠. Una vez que obtengo las ideas o soluciones 💡, las pego en Google AI Studio 🌎🤖 para generar código refinado y final ✨📝. El resultado es un código que generalmente funciona sin problemas y sin errores 🔄✅. Además, puedes considerar combinar Grok con OpenAI o alternar entre Google y OpenAI según la tarea específica 🔄🎯. Aunque Grok no tiene la enorme ventana de contexto de Google 🚫📊, a menudo está más actualizado con los avances recientes 📅📰.
En resumen: combinar las fortalezas únicas de estos modelos de IA puede mejorar significativamente tu eficiencia en programación 💪💻. Reconoce las capacidades de cada uno—ya sea la generación extensa de código de Google 🖥�, la mejor resolución de problemas de OpenAI 🛠�, o los conocimientos actualizados de Grok 📖—para optimizar tu flujo de trabajo de programación 🚀.
Siéntete libre de experimentar con estas herramientas y descubrir la combinación más efectiva para tus proyectos 🧪🔍.
#AgentesIA 🤖 #CodificaciónIA 💻 #OpenAIO3 💡 #GoogleAI 🌐 #Grok3 🧠
#2
OxygenBasic Examples / Re: Oxy Independent asm (old)
Last post by Frank Brübach - April 21, 2025, 10:25:32 PMhello charles, made this example works. just for fun..
usage: compile/compile and run binary32
output: two mesage boxes and display a command line..
Code Select
' only an experimental test, 21-04-2025, frank brübach
' and it works :-)
#file "indep1d.exe" 'independent -> no need for it
'------
'PROLOG
'======
' I have added -------------- //
def _epilog 'FOR EXE
====================
._end_
mov edi,eax 'hold exit value in edi
push 0 'place for exit code
mov eax,esp
push eax
call getModuleHandle
push eax
call GetExitCodeProcess
mov eax,edi 'return exit value in eax
push eax
call ExitProcess
._error_
push 0x30
"MISSING OR UNLOADABLE"
push eax
push ecx
push 0
call [ebx+472]
mov eax,0
jmp _end_
end def 'eplilog for exe
' I have added -------------- //
mbox "hello" ' only for test purpose
push ebx : push esi : push edi : push ebp
mov ebp,esp
sub esp,256
'--------------------------------------------------------------
'MOVE BOOTSTRAP PROCEDURE POINTERS TO RUNTIME LIBRARY POSITIONS
'==============================================================
'
'GET ABSOLUTE ADDRESSES
'
call fwd here
.here
pop eax
sub eax,here
mov ebx,eax : add ebx,bssdata
mov edi,eax : add edi,import_address_table
'
mbox "hello2" ' only for test purpose
'--------------------------------
'COPY BOOTSTRAP LIBRARY ADDRESSES
'================================
'
mov eax,[edi+00] : mov [ebx+024],eax 'LoadLibrary
mov eax,[edi+04] : mov [ebx+040],eax 'GetProcAddress
mov eax,[edi+08] : mov [ebx+032],eax 'FreeLibrary
mov eax,[edi+12] : mov [ebx+440],eax 'GetModuleHandle
mov eax,[edi+16] : mov [ebx+448],eax 'GetGetCommandLine
mov eax,[edi+20] : mov [ebx+456],eax 'GetExitProcess
mov eax,[edi+24] : mov [ebx+464],eax 'ExitProcess
mov eax,[edi+28] : mov [ebx+480],eax 'CreateFileA
mov eax,[edi+32] : mov [ebx+488],eax 'Readfile
mov eax,[edi+36] : mov [ebx+496],eax 'CloseHandle
mov eax,[edi+44] : mov [ebx+472],eax 'MessageBoxA
'=============
jmp fwd endlib
'=============
mbox "hello3" ' only for test purpose
'-----------
TestMessage:
'===========
push 0
"title dolphin"
push eax
"message ocean"
push eax
push 0
call [ebx+472]
ret
'------------------
DisplayCommandLine:
'==================
push 0
"COMMAND LINE"
push eax
call [ebx+448]
push eax
push 0
call [ebx+472]
ret
'-------
Message:
'=======
pop edi
call [ebx+472]
push edi
ret 16
'======
endlib:
'======
'-----
'TESTS
'=====
call TestMessage
call DisplayCommandLine
call Message 0,"message oxygen","title o2h",1
mbox "ok4"
'------
'EPILOG
'======
endprog:
mov esp,ebp : pop ebp : pop edi : pop esi : pop ebx
ret
' ends
#3
OxygenBasic Examples / Re: Oxy Independent asm (old)
Last post by Charles Pegge - April 20, 2025, 11:35:06 PMIt is a small part of RTL32, and does not mean very much without the PE file headers specified in inc\self\hdrs.inc. It is a small set of Windows API functions required to initiate Oxygen.dll before the remaining Runtime functions are constructed.
Excellent weather for Easter day, But very quiet down here. I heard a woodpecker in the distance, which is a good sign for our local ecology.
Excellent weather for Easter day, But very quiet down here. I heard a woodpecker in the distance, which is a good sign for our local ecology.
#4
OxygenBasic Examples / Re: Oxy Independent asm (old)
Last post by Frank Brübach - April 20, 2025, 08:01:13 PMYes thanks Charles I know :-)
Ok.. How this old example can translate to actual rtl32 inc File? Or thats Nonsens value? Thanks in advance nice Happy easter Party .. WE Had Sunny Times in Germany and all is full of blossom plants and tress
Ok.. How this old example can translate to actual rtl32 inc File? Or thats Nonsens value? Thanks in advance nice Happy easter Party .. WE Had Sunny Times in Germany and all is full of blossom plants and tress
#5
microA Interpreter / Re: micro(A) progies
Last post by Zlatko Vid - April 20, 2025, 09:32:08 AMWindows X 11 Lite Optimum 24H2
just installed on my Thinkpad w530
works excellent... [Image: grin.png] so far ,,,
Happy Easter !!!!
just installed on my Thinkpad w530
works excellent... [Image: grin.png] so far ,,,
Happy Easter !!!!
#6
OxygenBasic Examples / Re: Oxy Independent asm (old)
Last post by Charles Pegge - April 20, 2025, 07:02:38 AMHi Frank,
That looks like the beginnings of the 32bit run-time library: RTL32.inc. Before that, Oxygen binaries were dependent on including oxygen.dll with every app.
That looks like the beginnings of the 32bit run-time library: RTL32.inc. Before that, Oxygen binaries were dependent on including oxygen.dll with every app.
#7
OxygenBasic Examples / Class example: fleximembers
Last post by Frank Brübach - April 19, 2025, 11:39:20 PMCode Select
' Simple implementation for making runtime members
' - supports integer, float and string types.
'===================
class myfleximembers
'===================
indexbase 0
bstring buf, *varl
sys dp,en
method addVar(string name,dat)
sys le=len buf
if dp+16>le then
buf+=nuls 0x100 : le+=0x100 :
end if
@varl=?buf
varl[en]=name
varl[en+1]=dat
dp+=2*sizeof sys
en+=2 'next slot
end method
method find(string name) as sys
sys i
for i=0 to <en step 2
if name=varl[i] then return i+1
next
end method
method vars(string name) as string
sys f=find(name)
if f then return varl[f]
end method
method VarF(string name) as double
return vars(name)
end method
method VarI(string name) as sys
return vars(name)
end method
method vars(string name,dat)
bstring varl at buf
sys f=find(name)
if f then varl[f]=dat
end method
method delete()
sys i
sys v at buf
for i=0 to <en
freememory v[i]
next
freememory ?buf
? buf=0 : en=0 : dp=0
end method
end class
' demo test
myfleximembers z
z.addVar "p",5
z.addVar "q",4.5
z.addVar "r","123456"
print z.Vars("q")+z.vars("q") 'result 4.254.25
print z.Varf("q")+z.varf("q") 'result 9
print z.VarI("q")+z.vari("q") 'result 10
z.delete
' ends
#8
OxygenBasic Examples / Oxy Independent asm (old)
Last post by Frank Brübach - April 19, 2025, 10:41:28 PMhello charles, I have found this old o2 example. looks interesting.
how to use it? that's possible to make a message to print any output?
thanks, frank
Code Select
'
' This is an Oxygen-independent program - the bare bones without a proper run time library.
' It has the minimal set of API calls to make it usable.
' This program cannot be run directly - it has to be compiled first. The EXE file size is 3.5k.
' You can use it for simple Asm with stack-based memory. There is no built in code
' for memory management or "bind" function.
' Charles
'
#file "indep.exe" independent
'------
'PROLOG
'======
push ebx : push esi : push edi : push ebp
mov ebp,esp
sub esp,256
'--------------------------------------------------------------
'MOVE BOOTSTRAP PROCEDURE POINTERS TO RUNTIME LIBRARY POSITIONS
'==============================================================
'
'GET ABSOLUTE ADDRESSES
'
call fwd here
.here
pop eax
sub eax,here
mov ebx,eax : add ebx,bssdata
mov edi,eax : add edi,import_address_table
'
'--------------------------------
'COPY BOOTSTRAP LIBRARY ADDRESSES
'================================
'
mov eax,[edi+00] : mov [ebx+024],eax 'LoadLibrary
mov eax,[edi+04] : mov [ebx+040],eax 'GetProcAddress
mov eax,[edi+08] : mov [ebx+032],eax 'FreeLibrary
mov eax,[edi+12] : mov [ebx+440],eax 'GetModuleHandle
mov eax,[edi+16] : mov [ebx+448],eax 'GetGetCommandLine
mov eax,[edi+20] : mov [ebx+456],eax 'GetExitProcess
mov eax,[edi+24] : mov [ebx+464],eax 'ExitProcess
mov eax,[edi+28] : mov [ebx+480],eax 'CreateFileA
mov eax,[edi+32] : mov [ebx+488],eax 'Readfile
mov eax,[edi+36] : mov [ebx+496],eax 'CloseHandle
mov eax,[edi+44] : mov [ebx+472],eax 'MessageBoxA
'=============
jmp fwd endlib
'=============
'-----------
TestMessage:
'===========
push 0
"title"
push eax
"message"
push eax
push 0
call [ebx+472]
ret
'------------------
DisplayCommandLine:
'==================
push 0
"COMMAND LINE"
push eax
call [ebx+448]
push eax
push 0
call [ebx+472]
ret
'-------
Message:
'=======
pop edi
call [ebx+472]
push edi
ret 16
'======
endlib:
'======
'-----
'TESTS
'=====
'call TestMessage
'call DisplayCommandLine
call Message 0,"message","title",1
'------
'EPILOG
'======
endprog:
mov esp,ebp : pop ebp : pop edi : pop esi : pop ebx
ret
'print "ok"
#9
OxygenBasic Examples / Re: Question about MAKEINTRESO...
Last post by Charles Pegge - April 19, 2025, 04:31:34 PMThis is a quirk of early Windows where some functions read low value params as literal integers but also interpret high values as string pointers. It might be an elegant trick but also a source of confusion when some functions can do this and others cannot.
#10
OxygenBasic Examples / Re: Question about MAKEINTRESO...
Last post by Roland Stowasser - April 19, 2025, 04:04:59 PMThank you Charles. The function worked perfectly for me. It also worked for function LoadIcon in the WNDCLASSEX wc structure: Somehow this looks a bit like magic to me. MAKEINTRESOURCE makes an int 1000 to a string "#1000" which then can be used like a number in e.g. function LoadImage. I think something is happening behind the scenes?