Uploaded new version on mentors.
> >I knew what I did, but feel free to disagree with my reasoning.
> >
> >fgetty-login.diff essentially just renames /bin/login1 to /bin/fgetty-login.
> <I find it confusing, when binaries are renamed. Either way, /bin/login{1,2}
> >are never invoked directly, so I considered installing them into /lib/fgetty.
> >It would eliminate need in manpage, which is barely useful. WDYT?
Installed `login1' and `login2' into /lib/fgetty/. Looks nice to
me. Dropped manpage.
> > Yes, just tried, `getty' from util-linux
> >issues warning, not error in case of ro /dev, but can anyone explain me
> >why it is good thing?
> it might not be the same on Debian derivatives, on chroots, on
> minimal installations, on raspbian and so on. in my opinion this
> one is still needed (and should be upstreamed).
>
> I never had to mount /dev in ro, but I tried in a chroot and it seems
> possible.
>
> Some particular installations might have this, for security or whatever else.
>
> How do you feel about reverting?
I am not fond of maintaining code for some strage setup, not
supported by upstream. I tried to search web, and I found no information
about read-only /dev. Read-only /, /etc/, /usr -- sure, but not /dev.
Adding Pape in CC, he should be able clarify issue. I will revert, if
someone bug an report about his read-only /dev.
> >3 of 4 binaries are statically linked with no debug information. So
> >the only debug symbols for checkpassword(1). Since `$(wc -l
> >checkpassword.c) = 93` I see little use of debug package.
> ok, but shouldn't the new system provide "empty" debug packages then, or
> maybe even better, understand and don't provide them?
Probably. But it is question to debhelper team. I can't remove "empty"
debug ELF's manually, since their name are some hashes.
> >Sure. Will do better next time.
I documented my position on ro-dev.diff and debug packages in changelog.
--
Accept: text/plain, text/x-diff
Accept-Language: eo,en,ru
X-Keep-In-CC: yes
X-Web-Site: sinsekvu.github.io
