PureBasic Forums - English

Hello everyone !

- The final is out, you can grab it on your personal account. Thanks a lot to everyone during this bug-fix marathon

- beta 5 is out, with some more bug fixes, don't hesitate to try it out ! Updated GCC on Windows to 12.2.0 version.

Bugfixes for this version:

• Windows: [Done] WebGadget on PB 5.31 and 5.30 to OLD !
• Windows: [Done] PureBasic WebGadget documentation example throws errors
• Windows: [Done] WebGadget PB6.0 Error
• Windows: [Done] #PB_ProcessPureBasicEvents is not listed in "PureBasic Constants"
• Windows: [Done] Regression PB5.61 Binary not (~) doesn't works with 64bits constants
• Windows: [Done] PB 5.62-5.72 Interfaces inside structures IMA
• Windows: [Done] Assembler Error
• Windows: [Done] Load PNG Image bug still in PB5.41 LTS
• Windows: [Done] PB 5.42 Crash loading a PNG image to place in the toolbar
• Windows: [Done] PB5.40LTS PB sometimes crashes while opening images
• Windows: [Done] Crash on certain PNG image when TGA decoder is loaded.
• Windows: [Done] IMA on certain corrupted PNG image files.
• Windows: [Done] LoadImage Crashes Program
• Windows: [Done] x64 int "ext"
• C Backend: [Done] PB6.00x86/6.01b3x86 Problem with bitwise operators and quads
• C Backend: [Done] IMA on break
• C Backend: [Done] Different bug message at compilation 2
• C Backend: [Done] Compiler error for ProcessIdToSessionId

This list was created automatically. All updated bug threads from the date of the last final version have been determined.

- beta 4 is out, with some more fixes. This will be the last beta for now, so be sure to test it properly so we can have a good final release !
Note: The optimizer needs to be manually enabled for ASM backend as well to have any effect.
The event handling code has been modified, so if you find something odd in event propagation, please report it.

Bugfixes for this version:

• Windows: [Done] Window 10 (1903) and PB-v5.7x / Floating-Registers failure
• Windows: [Done] Nested macros, string lost along the way
• Windows: [Done] PB5.70b1 single line macro resolving error
• Windows: [Done] Malformed Assert() statement
• Windows: [Done] Stack alignement bug on X64 ?
• Windows: [Done] toolbarbutton toggle
• Windows: [Done] Pointer bug
• Windows: [Done] ComboBoxGadget's State bug
• Windows: [Done] PB 5.70 editable ComboBox() with icon wrong GetGadgetState()
• Windows: [Done] Error message with fileformat LF - Includefile
• Windows: [Done] Purifier gives false positive errors
• Windows: [Done] EnablePurifier + ReAllocateMemory
• Windows: [Done] Bug RemoveKeyboardShortcut(0, #PB_Shortcut_Tab)
• Windows: [Done] ExplorerTreeGadget and #PB_EventType_RightDoubleClick
• Windows: [Done] Disabled ButtonImageGadget() not shown OK with JPG files
• Windows: [Done] KeyboardInkey() can return two characters
• Windows: [Done] KeyboardInkey() doesn't return unicode
• Windows: [Done] ExaminePack()+PackEntryName(): problems with string encoding
• Windows: [Done] OpenDatabase fails when using "scram-sha-256" authentication
• Windows: [Done] GetActiveGadget fails with ComboBoxGadgets
• Windows: [Done] OpenNetworkConnection() with IPv6 fails
• Windows: [Done] Wrong gadget size with dialog library?
• Windows: [Done] Variable already declared With another type
• Windows: [Done] FormatXML() removes spaces inside comments
• Windows: [Done] Bind and unbind gadgets
• Windows: [Done] /PREPROCESS and IncludePath problem
• Windows: [Done] *Pointer = @Value adds unnecessary asm instructions
• Windows: [Done] asm error
• Windows: [Done] FormatDate with invalid date
• Windows: [Done] Dialog Lib IMA when rowspan > 68
• Windows: [Done] Annoying bug with editor and high dpi settings
• Windows: [Done] Strange ASM-Compiler message with PB-6.01B3 LTS x64
• Windows: [Done] 6.01b3 - Encoding problem in debugger Window
• C Backend: [Done] PB6.01b3 x64 - Self referencing structure issue :
• C Backend: [Done] 6.01b3 - Map not initialized in C Backend

This list was created automatically. All updated bug threads from the date of the last final version have been determined.

- beta 3 is out, with some more bug fixes to test !

Bugfixes for this version:

• Windows: [Done] DesktopScaledX / Y not working?
• Windows: [Done] PB 5.42 Window loses background colour
• Windows: [Done] ComboBoxGadgets do not render correctly on Windows 11 when a window background color is set
• Windows: [Done] SmartWindowRefresh() broken on Win 11
• Windows: [Done] Text truncated with #PB_Gadget_RequiredSize
• Windows: [Done] resizing ExplorerComboGadget and DPI scaling
• Windows: [Done] SpinGadget in Dialog has no natural size
• Windows: [Done] CreatePopupImageMenu dpi unaware
• Windows: [Done] CreatePopupImageMenu ignores dpi setting
• Windows: [Done] 6.00 Beta 10: Bigger Images in PopupMenus are displayed at full size when disabled
• Windows: [Done] PB 5.73 LTS x86 ComboBoxGadget not correctly auto-sizing
• Windows: [Done] TreeGadget SetGadgetState -1 not working
• Windows: [Done] ClipOutput() + FillArea() corrupts memory
• Windows: [Done] DocMaker does not work in PureBasic 6.00 Beta 6 (Windows - x64)
• Windows: [Done] SendNetwork...() fails with UDP via IPv6
• Windows: [Done] PB 5.50 SendNetworkData using UDP over IPv6 does not work
• Windows: [Done] PB 5.50 Client/server with UDP + IPv6 are bugged
• Windows: [Done] Network GetClientIP UDP IPv6
• Windows: [Done] ExtractXMLStructure and memory corruption
• Windows: [Done] SendMail Outlook/Office365 broken in 5.71 (works in 5.62!)
• Windows: [Done] strip.exe generates wrong path
• C Backend: [Done] PB6.01b1 C-backend reduced speed of app to half
• C Backend: [Done] PB6.01 b1 x86 - Multiply not commutative
• IDE: [Done] 6.01B1 Autocomplete shows wrong entries for module [FIX INSIDE ?]
• IDE: [Done] 5.30B7 Autocomplete shows CompilerIf as a structure's field [FIX INSIDE ?]

This list was created automatically. All updated bug threads from the date of the last final version have been determined.

- beta 2 is available, with some more bug fixes, don't hesitate to test it to help us having a reliable release !

Bugfixes for this version:

• Windows: [Done] Dialog bug when using frames in high DPI?
• Windows: [Done] strip.exe generates wrong path
• Windows: [Done] bug listicongadget()
• Windows: [Done] RemoveGadgetItem on ComboBoxGadget Bug
• Windows: [Done] PB 6.00 LTS x86 Weird problem with RemoveGadgetItem and SortStructuredArray
• Windows: [Done] Debug issue (was: PeekS() issue)
• Windows: [Done] Email code works in PB 5.61 but not 5.70 or 5.71
• Windows: [Done] Problem with SMTP Mailer an PB 5.72
• Windows: [Done] SendMail doesn't work for Versions 5.7x and 6.00 Beta
• Windows: [Done] SendMail Outlook/Office365 broken in 5.71 (works in 5.62!)
• Windows: [Done] Splitter sometimes not working
• Windows: [Done] DPI in Editor (again)
• Windows: [Done] [AES] StartAESCipher error
• Windows: [Done] OpenDatabase() has a problem with space characters in database names [all OS]
• Windows: [Done] CalendarGadget and GetGadgetText
• Windows: [Done] HTTPRequest() with #PB_HTTP_Asynchronous flag
• Windows: [Done] Compiler Crash Bug pbide 6 x64
• Windows: [Done] What is #PB_EventType_SizeItem for?
• Windows: [Done] OpenWindowedScreen and window resize
• Windows: [Done] Ellipses with negative RadiusX Not Drawn
• Windows: [Done] Can't set EditorGadget to #CRLF$ alone
• Windows: [Done] procedure parameters matching Define
• Windows: [Done] Bug array size in x64
• Windows: [Done] 5.51 Label address arithmetic bug
• C Backend: [Done] Win x64 v6.0 LTS - Optimizer crippled
• C Backend: [Done] PB 6.00 Beta 8 - Compiler crash with large sources

This list was created automatically. All updated bug threads from the date of the last final version have been determined.

The bug fix version 6.01 beta 1 is available on your online account. It includes many old timer and C backend related bugs. Feel free to test it as much as possible as it could have some regressions due to several big internal changes.

Bugfixes for this version:

• Windows: [Done] 5.51 Label address arithmetic bug
• Windows: [Done] Bug or not bug
• Windows: [Done] Arithmetic Compiler Bug
• Windows: [Done] 4.60 RC1 Val() in Static Array Index
• Windows: [Done] PB 4.30 - 5.61 x64 - NaN
• Windows: [Done] Procedure stack corruption - PB x86 ASM - Quad-Double-String handling
• Windows: [Done] Bug x64 assigning FPU result to a field in a structure
• Windows: [Done] x64 Structured integer and quad variables behave differently
• Windows: [Done] PBx86 FormatNumber() with procedure calls
• Windows: [Done] Bug array size in x64
• Windows: [Done] PB 5.11 (x86) - Pointer plus Quad
• Windows: [Done] bug? i cant make sense of this little one
• Windows: [Done] Floating point recursion - Is this a bug?
• Windows: [Done] DatabaseColumnType() Reporting Wrong Type in Postgres
• Windows: [Done] Direct debug of a double cacluation result in a float
• Windows: [Done] Strange behavior with Round(), Val() and .l variable type
• Windows: [Done] 4.50 - 4.60 Implicit cast from float to integer
• Windows: [Done] Error comapring float to int
• Windows: [Done] Comparing integer and float (or double) variables
• Windows: [Done] Problems of using floats with IF
• Windows: [Done] If a$="N" Or a$="A" And a<>1 why so?
• Windows: [Done] 5.61 - Bool() returns numbers <> 1
• Windows: [Done] 4.51 x86 weird result of the If comparison
• Windows: [Done] 5.40b1(and all other) Boolean operators
• Windows: [Done] 4.31 Boolean operators are right associative
• Windows: [Done] PB511 AND OR expression, wrong result
• Windows: [Done] Faulty And/Or Processing
• Windows: [Done] Inconsistent application of banker's rounding
• Windows: [Done] PB 5.31 - Wrong ASM generated for sign extension
• Linux: [Done] Some differences between Windows and Linux
• MacOS: [Done] Openscreen() always returns 0 on Mac M1
• MacOS: [Done] InitSound() Fails on Big Sur
• C Backend: [Done] Win x86 BITAND casting
• C Backend: [Done] Inconsistent output between PB compilers
• C Backend: [Done] PB 6.00 - Compiler error
• C Backend: [Done] Output difference - float vs double
• C Backend: [Done] Quad not always commutative
• C Backend: [Done] LF missing in file.pp pbcompilerc --preprocess
• C Backend: [Done] Bug:Errror "expected ";"before DBG_Check"
• C Backend: [Done] Assembler Error in 6.00 beta 10 Win x64.
• C Backend: [Done] copy structure bug
• C Backend: [Done] macOS: PB-IDE Build Error
• C Backend: [Done] Integer bit shift in x64
• C Backend: [Done] macOS IDE Start Bug
• C Backend: [Done] DPI
• C Backend: [Done] OnErrorCall not working for Console

This list was created automatically. All updated bug threads from the date of the last final version have been determined.

Have Fun,

The Fantaisie Software Team

Thanks for your efforts!

Do you implement test cases for all fixed bugs to avoid regressions in the future?

NicTheQuick wrote: Fri Jan 20, 2023 1:52 pm Do you implement test cases for all fixed bugs to avoid regressions in the future?

Yes, we always add an automatic test for every bug fixed (unless it's GUI stuff which are harder to do).

@Fred

Does it support GCC -O3?

Is GCC compiled code faster than in 6.00?

Thank you!

It doesn't support O3 and will probably never will. O3 brings so much aggressive optimizations, than even Linus doesn't want to use it for the Linux kernel. Here is an interesting read about this: https://www.phoronix.com/news/No-O3-For-Linux-Kernel

That said, 6.01 should be much faster than 6.00 due to volatile rework, but it still needs to be confirmed working as expected.

Fred wrote: Fri Jan 20, 2023 2:04 pm It doesn't support O3 and will probably never will. O3 brings so much aggressive optimizations, than even Linus doesn't want to use it for the Linux kernel. Here is an interesting read about this: https://www.phoronix.com/news/No-O3-For-Linux-Kernel

That said, 6.01 should be much faster than 6.00 due to volatile rework, but it still needs to be confirmed working as expected.

Thank you!

I didn't know about the -O3 issues.

+++
Big hank you Fred!

Thanks for making PB better and better.
Great work Fred and team !

Thx from me too.

The list also shows bug which seem to be present since v4.5 and earlier. Wow.

I normally use -O2 with gcc but in rare occasions -O3 is significantly faster, but most often the difference between -O2 an -O3 is small
but why not give the user the option to choose the O level?
or for that matter any other gcc options

Many thanks Fred for the never ending effort to make PB reach perfection !
What a great loving dev product !
I'm so happy, I have been using it for 10 years now !
PS : I use it to make all my tools (at light speed!) for retro game development !

jack wrote: Fri Jan 20, 2023 4:05 pm I normally use -O2 with gcc but in rare occasions -O3 is significantly faster, but most often the difference between -O2 an -O3 is small
but why not give the user the option to choose the O level?
or for that matter any other gcc options

Because if add the option to select o3 ppl will use it and bug tracking in non real bug will be start and create a mess ..i guess

Thank you Fred & Team.

Impressive to see PB 4.30, 4.50, 4.60 RC1, 4.60,...
With old bugs fixed, some dating back to 2011. I didn't know anything about PureBasic, even the name at that time.
Wow, What a long way, congratulations for the quality of the work done and for maintaining this great Language, so powerful, light and fast, again and again.