+1 maintenance

Danilo Egea Gondolfo danilo.egea.gondolfo at canonical.com
Fri Mar 1 19:46:31 UTC 2024


Hello,

I had some Netplan related tasks to finish before FF so my +1 maintenance
shift wasn't super efficient.

For awareness, I'm seeing lots of similar FTBFS issues:

1) FTBFS due to missing symbols in symbol files. Some of them are due to
only two symbols, strlcpy and strlcat. Those functions were added in glibc
2.38 and started to
show up now during rebuilds.

2) FTBFS due to custom implementations of strlcpy and strlcat and the new
glibc.

I suppose Debian is not seeing it yet because glibc 2.38 is still in
experimental there.

3) FTBFS due to unsatisfiable dependencies. Probably caused by the ongoing
time_t work.

4) FTBFS due to _FORTIFY_SOURCE redefinition. It started to happen probably
because we bumped it up to 3 in Ubuntu and it's still 2 in Debian.

Things I worked on:

fetchmail: FTBFS due to OpenSSL version (LP: #2052964). fetchmail is
compatible with the OpenSSL available in Noble so I dropped the dependency
from >= 3.1.x to >= 3.0.x.
OpenSSL 3.0 is the version we are going to ship with Noble.

libhtp2: FTBFS on Noble on all archs (LP: #2055051). The symbols file was
missing some new symbols and the build is failing on Ubuntu. It's blocking
suricata.

btas: btas is causing a FTBFS in tiledarray. I submitted a no change
rebuild (LP: #2055063) due to the new libboost1.83. btas installs a cmake
file with the full path of BoostConfig.cmake and it's pointing to the old
file from libboost1.74.

android-platform-art: FTBFS due to new glibc strlcpy symbol (LP: #2055399).
I've created a little patch to only compile the package’s own
implementation of strlcpy if glibc is older than 2.38.

beep: FTBFS due to _FORTIFY_SOURCE redefinition (LP: #2055689)

intel-graphics-compiler: FTBFS due to _FORTIFY_SOURCE redefinition. I
prepared a patch but the maintainer was quicker and fixed it in Debian.
Although, the solution seems to not be ideal for Ubuntu. The change
basically undefines _FORTIFY_SOURCE so when it shows up again it doesn't
fail with a redefinition error. Although, on Ubuntu, this is what happens
effectively: gcc -D_FORTIFY_SOURCE=3 -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2
so the default value of 3 from Ubuntu is overwritten back to 2.

libdumbnet: FTBFS due to the symbols file and strlcpy/strlcat symbols with
the new glibc (LP: #2055693).

libwww-mechanize-perl: merged from debian (LP: #2055713)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.ubuntu.com/archives/ubuntu-devel/attachments/20240301/1e6d6f8b/attachment.html>


More information about the ubuntu-devel mailing list