Tiny Core 17.0 Alpha 1 Testing

[aus9]

OK managed to photo with handheld camera the K messages scrolling by
Image names are left for left side of monitor and right

right is
https://i.postimg.cc/P5JZHnPn/right.png 
To save you clicking it --reads
/sbin/modprobe -bv acpi:PNP0C02 [changing number in this field]

left side is
https://i.postimg.cc/B6RsjDCn/left.png
to save you clicking it --reads SNIP numbers
udevd timeout: killing /sbin/modprobe as per right image

[aus9]

Now I attempt to look at ACL as per link
https://www.linuxbash.sh/post/enabling-and-managing-acls-on-filesystems

my system is on /dev/sdb1 and

Code: [Select]

tune2fs -l /dev/sdb1 | grep "Default mount options:"
Default mount options:    user_xattr acl
So next I edit my fstab so it reads
/dev/sdb1       /mnt/sdb1       ext4     users,exec,acl    0 1
and full reboot  but dmesg still shows Failed to apply ACL: Operation not supported

I am not a coder but maybe its my rootfs I need to check?

Code: [Select]

getfacl /dev
getfacl: Removing leading '/' from absolute path names
# file: dev
# owner: root
# group: staff
user::rwx
group::rwx
other::r-x
beyond my pay scale at this point

[Juanito]

xtables-addons-6.12.11-tinycore64.tcz  and  xtables-addons-6.12.11-tinycore.tcz
don't belong in the 17.x repos, do they?

Correct - removed, thanks

[andyj]

Hi Juanito
xtables-addons-6.12.11-tinycore64.tcz  and  xtables-addons-6.12.11-tinycore.tcz
don't belong in the 17.x repos, do they?

I'm working on those.

[andyj]

This little tidbit from the GCC 15 changes page is proving to be a challenge for several of the extensions I maintain:

To aid the transition to C23, various diagnostics have been enhanced to clarify type errors such as incompatible function types, incorrect argument counts, and those involving bool.

I know that a lot of extensions are being copied from earlier TCL releases because they seem to run, but could they be complied on the newest release? Should this be one of the tests?

[andyj]

Does this mean I have to recompile mariadb:

lto1: fatal error: bytecode stream in file '/tmp/tcloop/mariadb-12.1-dev/usr/local/mysql/lib/libmariadb.a' generated with LTO version 14.0 instead of the expected 15.1                                                                 

[aus9]

I confirm that python3.14 works for  updated yt-dlp I am submitting

[aus9]

I know its early days....I can not see python3.14-pip and can work around it but python complained

Code: [Select]

python3.14 setup.py install --prefix=/usr/local --root=/tmp/$P
SNIP
Please avoid running ``setup.py`` directly.  Instead, use pypa/build, pypa/installer or other   standards-based tools.and gave a link
https://blog.ganssle.io/articles/2021/10/setup-py-deprecated.html
SNIP  setup.py install -> pip install

thanks for reading

[Juanito]

I can not see python3.14-pip and can work around it but python complained

pip is provided by python3.14

[aus9]

oops

[Juanito]

I know that a lot of extensions are being copied from earlier TCL releases because they seem to run, but could they be complied on the newest release? Should this be one of the tests?

There are usually some problems with compiling with a new major version of gcc, but this is gcc-15.2.0, so it's been out a while, so you would think that most problems have been ironed out.

For the issues related to c23, you could try building with an earlier version of the c/c++ standards specified?

[Juanito]

Does this mean I have to recompile mariadb:

lto1: fatal error: bytecode stream in file '/tmp/tcloop/mariadb-12.1-dev/usr/local/mysql/lib/libmariadb.a' generated with LTO version 14.0 instead of the expected 15.1                                                                 

That's the first time I've seen such an error - are you using python3.14? Would using "-fno-lto" help?

As an aside, when I compile the few extensions I maintain that include static libraries, I make a second pass without lto specifically for the static libs, because otherwise they will be massively inflated in size by lto.

[andyj]

For the issues related to c23, you could try building with an earlier version of the c/c++ standards specified?

I did set some to use -std=gnuc17 which solved some problems. However, I'm getting this with Postgresql 16+ (earlier versions don't have this problem, and none of them did in TCL 16):

/usr/local/bin/msgfmt: Cannot convert from "ISO-8859-1" to "UTF-8". msgfmt relies on iconv(). This version was built without icon

Did something change that could affect this?

[andyj]

That's the first time I've seen such an error - are you using python3.14? Would using "-fno-lto" help?

As an aside, when I compile the few extensions I maintain that include static libraries, I make a second pass without lto specifically for the static libs, because otherwise they will be massively inflated in size by lto.

Recompiling mariadb fixed the issue. Regarding the static libraries, mariadb uses cmake. I looked through the various cmakefile's and I didn't find anything that I thought looked like it would have the same affect as --disable-static would.

As for python, I don't use it intentionally so I don't know what version is being used.

[Juanito]

/usr/local/bin/msgfmt: Cannot convert from "ISO-8859-1" to "UTF-8". msgfmt relies on iconv(). This version was built without iconv

Did something change that could affect this?

gettext is updated in tc-17, but I have seen this error from time to time in previous versions.

Sometimes loading glibc_apps/glibc_gconv/glibc_i18n_locale has solved the problem and once I removed the test for iconv in a configure script and the resulting app worked...