FPC 3.2.4-rc1 available

[AlexTP]

Pls merge fix mentioned here (I dont have URL to commit):
https://forum.lazarus.freepascal.org/index.php/topic,74216.msg587438.html#msg587438

It was related to this problem, already fixed in Git main:

fcl-res: COFF resource writer emits uninitialized NumAuxSymbol, producing a malformed object file

Component: Packages / fcl-res
Target: all COFF targets (win32/win64/wince/arm), reproduced on x86_64-win64
Severity: minor in normal use, but causes a hard linker crash with `-Xe -XX`

[marcov]

As said in that thread, I can see the fix in fixes_3_2.   I think it was in the batch of fixes  1,2 months ago, as it was marked merged, but still visible in the mergelogs app that I use during mass merges.

[BSaidus]

I saw the 3.2.4-rc2 in gitlab. good news 

[440bx]

I saw the 3.2.4-rc2 in gitlab. good news 

That is good news. 

Hopefully the formal release isn't too far away.

[dbannon]

I saw the 3.2.4-rc2 in gitlab. good news 

yes, I noticed rc2 appeared too. But it has a confusing sort of version number.

With the tagged release_3_2_4_rc1 the compiler identifies itself as 3.2.3 when, for example, its speaking to the fpc.cfg file about where units might be.  I assume that comes from the compiler->version.pas file that contains -

Code: [Select]

interface
    const
       { version string }
       version_nr = '3';
       release_nr = '2';
       patch_nr   = '3';
       minorpatch = '';
Good. But rc2, looks quite different, its compiler->version.pas says -

Code: [Select]

interface
    const
       { version string }
       version_nr = '3';
       release_nr = '2';
       patch_nr   = '4';
       minorpatch = '-rc1';
and sure enough, it believes it is 3.2.4 when, again, speaking to fpc.cfg.  Maybe this does not matter, it did upset my scripts that I use to make cross compilers, easy fixed but one or the other appears wrong ?

Davo

[Thaddy]

The 3.2.4 (without RC's) branch is built from fixes, not from 3.2.4RC1/RC2, and that is the correct order,  so that is a small bug indeed. Report it ASAP so it can be fixed.

[dbannon]

Thanks Thaddy.

Both release candidates are tags not branches. Or are they tags along a branch ?

Anyway, I assume one is correct, one is wrong. If its rc1 that is wrong, little point in adding to the noise. On the other hand, if its rc2, its probably worth while, given it has not, yet been announced.

Its something of a problem when 3.2.4 is released however, both the release and rc2 expect to live under a 3.2.4 directory. Hmm...

Davo

[Thaddy]

@dbannon

I trust you reported it? or should I? Amongst probably others, Michael is preparing a release, if I understood correctly from the commits. These tiny bits help...

[marcov]

- In general, releases or RCs are branched from fixes, and after branching, the version is updated. This means that fixes stays at the old version till the release (3.2.4 final) is done, and then will change to 3.2.5

- Tags are implemented as branches in both GIT and SVN.   CVS had a separation that still persists in the naming.

- 3.2.4RC2 is being built, and mostly considered done. It is not announced yet. There are still a few weeks to evaluate the RC2 after it is announced before the final, where high prio fixes can be done.

[dbannon]

OK, I have logged this. Still don't feel comfortable about it, its a policy problem, not a technical one.

https://gitlab.com/freepascal.org/fpc/source/-/work_items/41893

Davo