Regressions in keeping minbase variant minimal since buster
Out of interest I've checked the state of sid vs buster debootstrap
--variant=minbase chroots to see if it's been growing while nobody has been
paying attention. Apparently we have a couple of regressions. Sharing this in
case someone else is interested in the result (or atleast hoping to wake
someones interest so they fix it so I can be lazy ;-P).
First here's a diff between buster-minbase and sid-minbase package list:
$ diff -u buster-minbase-pkg.list sid-minbase-pkg.list | grep '^[+-]'
--- buster-minbase-pkg.list 2020-01-01 13:53:54.528579149 +0100
+++ sid-minbase-pkg.list 2020-01-01 13:51:36.656253717 +0100
-gcc-8-base:amd64 install
+gcc-9-base:amd64 install
+libcrypt1:amd64 install
-libhogweed4:amd64 install
+libhogweed5:amd64 install
-libnettle6:amd64 install
+libnettle7:amd64 install
+libpcre2-8-0:amd64 install
+logsave install
+lsb-base install
My analysis of this follows:
Replacements (diff noise):
gcc-8-base -> gcc-9-base
libhogweed4 -> libhogweed5
libnettle6 -> libnettle7
Package splits (not really regressions):
libcrypt1 -> split from glibc (pulled in by libc6, login, passwd)
logsave -> split from e2fsprogs (pulled in by e2fsprogs - which is,
since buster, deinstallable!)
New (regressions):
libpcre2-8-0 pulled in by libselinux1
- several core packages have moved to pcre2, so this is basically an
unfinished libpcre(3) -> libpcre2 replacement.
- grep should also move over so libpcre3 is no longer needed. Already
reported in #907652
- (outside minbase there are however many other packages that will also
need porting, eg. glib. Thus it might be hard to avoid having both
PCRE libraries installed for a long time ahead.)
lsb-base pulled in by sysvinit-utils
- Already reported in #946399
- it might however finally be time to tackle the bigger task of making
sysvinit-utils non-essential, see #851747. I'm tempted but will post
details in a separate mail if I get motivated enough. Please reach
out if you're interested in helping out! I've given this plenty of
thought over the years so have a pretty good idea about it.
Regards,
Andreas Henriksson
Reply to: