Lazarus Release Candidate 2 of 3.0

[ermeneuta]

Issuing a make bigide under Debian 11 for the Rock 5A board, equipped with an ARM processor, works ok.
Doing the same under Armbian for the same board, fails as it is unable to resolve the two entry points :

__libc_csu_init
__libc_csu_fini

those entry points were present in glibc before 2.34. Starting from this release they were changed to

call_init
call_fini

Armbian has glibc 2.35, while Debian 11 has glibc 2.31
Apparently those two names are hardcoded somewhere in the build process of Lazarus.
Could it be possible to change that, in order to make possible a build of Lazarus even under Armbian ?

Thanks
-Fred

[dsiders]

Issuing a make bigide under Debian 11 for the Rock 5A board, equipped with an ARM processor, works ok.
Doing the same under Armbian for the same board, fails as it is unable to resolve the two entry points :

__libc_csu_init
__libc_csu_fini

those entry points were present in glibc before 2.34. Starting from this release they were changed to

call_init
call_fini

Armbian has glibc 2.35, while Debian 11 has glibc 2.31
Apparently those two names are hardcoded somewhere in the build process of Lazarus.
Could it be possible to change that, in order to make possible a build of Lazarus even under Armbian ?

Thanks
-Fred

An FPC bug report https://gitlab.com/freepascal.org/fpc/source/-/issues/39295 (closed) for this issue says it has been fixed in 3.3.1... but not yet merged into the fixes branch.

There has to be a better place than a release candidate announcement to discuss these things...

[ermeneuta]

There has to be a better place than a release candidate announcement to discuss these things...

Well, not knowing that the error was fpc-related, I (wrongly) supposed it was introduced in latest release candidate of Lazarus...

Sorry,
-Fred

[snorkel]

Not sure if this is a 3.0 issue but I didn't notice it until I started using RC2

Basically if you turn off the manifest DPI awareness the main menu control has a incomplete border on the top of the form, see screenshot.
I turn off the manifest DPI awareness as it does not look right, aka looks funny.

[WooBean]

..
Not my area of expertise, but I don't think line breaks are the issue.

Dotted unit names "casse.nfe" are a relatively new feature.... So could be (but don't know).
..
If you are using 3.0RC2 with a version of FPC other than 3.2.2 then make sure to mention this.
..

"Not my area of expertise" - you must be joking.

I was playing with "dotted units/identifiers" for a while.
Some progress has been done.
If interested - see at  https://gitlab.com/freepascal.org/lazarus/lazarus/-/issues/40596

[Gigatron]

Thanks to the Lazarus Team;

Gtr

[cpicanco]

Hi, just tried to compile a project and I am getting a "illegal qualifier" for a "Self.ClassName.Replace" call.

[cdbc]

Hi
From TObject:

class function ClassName : shortstring;

It could be that the stringhelper doesn't do shortstrings...
edit: use aStr:= copy(Classname,2); // 3.rd param implicits to rest of string
Regards Benny

[cpicanco]

I remember that someone fixed this in trunk. Is this helper not supposed to be released?

[Bart]

I remember that someone fixed this in trunk. Is this helper not supposed to be released?

If so, then it woul be fixed in fpc main (string helpers are part of fpc), so it may be in the next release of fpc.
This topic is about Lazarus relase.

Bart

[cpicanco]

I remember that someone fixed this in trunk. Is this helper not supposed to be released?

If so, then it woul be fixed in fpc main (string helpers are part of fpc), so it may be in the next release of fpc.
This topic is about Lazarus relase.

Bart

Ohhh, my bad. Sorry guys.

[DimTim]

I found these bugs in the trunk version, but I think they are also in RC2.

1) Hint sizes are calculated incorrectly, hint window's vertical size is too narrow;

2) The position of hints and menus for buttons on the right side of the screen or tray icons is also calculated incorrectly (for wide screens?  when DPI changed in Windows settings?)

System: Windows 10 Pro 22H2, hi-res monitor 2560x1440, scaling 150%.

[AlexTP]

Looks like a regression, pls don't release Lazarus yet.
https://gitlab.com/freepascal.org/lazarus/lazarus/-/issues/40615

[Martin_fr]

1) Hint sizes are calculated incorrectly, hint window's vertical size is too narrow;

2) The position of hints and menus for buttons on the right side of the screen or tray icons is also calculated incorrectly (for wide screens?  when DPI changed in Windows settings?)

I can't reproduce the hint size issue (I am on Win 10). Though there is a similar fixed issue https://gitlab.com/freepascal.org/lazarus/lazarus/-/issues/40565 (fixed after RC2).
However this issue was triggered by extremely long text in the hint (not the case in your screenshot).

Yet, it may still be, that its not the height, but the width that is calculated wrong. => The hint is may not be supposed to wrap the text. If a hint is calculated to narrow, then it can wrap, and the size won't be extended for that, because the wrapping was not expected.

---
About the hint position.

There is known issue, if you have 2 (or more screens) with different DPI settings. (Or maybe if you change settings, while the IDE is running).

If an IDE Window is focused on the screen with DPI setting "A", and you hover an IDE window on the screen with DPI setting "B", then the position will be wrong, as it is rescaled with the DPI settings from the focused window (and those don't apply).

[hukka]

1) Hint sizes are calculated incorrectly, hint window's vertical size is too narrow;

I believe this is a bug in Metadarkstyle instead as I'm hit with the same issue but it's fine if I remove the package.