[resending with correct list addr in CC]

Hello Rob,

David and I are looking at your design here at debcamp.

We noticed that the triggers spec[0] says "It is not defined in what
order triggers will run." We think this may invalidate your current
approach. Or did you already see this and account for it?

Thanks.

[0] /usr/share/doc/dpkg/spec/triggers.txt

--
Sean Whitton

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Sean Whitton <spwhitton@spwhitton.name> writes:

We noticed that the triggers spec[0] says "It is not defined in what
order triggers will run." We think this may invalidate your current approach. Or did you already see this and account for it?

I'll have to refresh, but from what I recall offhand, that might be a
problem. (And perhaps I missed/forgot that from the spec.)

And for what it's worth, when I left things last, I'd mostly been
reasoning from Manoj's last graph here (and the subset covered in
policy): https://people.debian.org/~srivasta/MaintainerScripts.html

In any case, if I remember correctly, I was under the impression that
they may respect dependency ordering at least to the extent that the
postinst configure does, and hence allow us to avoid having to handle
that ourselves (e.g. as we do now, not entirely satisfactorily, via
tsort). I believe I also did some testing in a VM, and the ordering was respected for some test packages I created, but of course that's not a
promise.

Thanks for taking a look.
--
Rob Browning
rlb @defaultvalue.org and @debian.org
GPG as of 2011-07-10 E6A9 DA3C C9FD 1FF8 C676 D2C4 C0F0 39E9 ED1B 597A
GPG as of 2002-11-03 14DD 432F AE39 534D B592 F9A0 25C8 D377 8C7E 73A4

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Hello,

On Tue 23 Jul 2024 at 08:18pm -05, Rob Browning wrote:

Sean Whitton <spwhitton@spwhitton.name> writes:

We noticed that the triggers spec[0] says "It is not defined in what
order triggers will run." We think this may invalidate your current
approach. Or did you already see this and account for it?

I'll have to refresh, but from what I recall offhand, that might be a problem. (And perhaps I missed/forgot that from the spec.)

And for what it's worth, when I left things last, I'd mostly been
reasoning from Manoj's last graph here (and the subset covered in
policy): https://people.debian.org/~srivasta/MaintainerScripts.html

In any case, if I remember correctly, I was under the impression that
they may respect dependency ordering at least to the extent that the
postinst configure does, and hence allow us to avoid having to handle
that ourselves (e.g. as we do now, not entirely satisfactorily, via
tsort). I believe I also did some testing in a VM, and the ordering was respected for some test packages I created, but of course that's not a promise.

Thanks. When David and I talked about it, we thought that we shouldn't
rely on ordering that's not guaranteed by the triggers spec -- though
possibly the spec has fallen out-of-date with the implementation.

Anyway, it seems like some more thought is required for the
emacsen-common rewrite, so we are going to go ahead and move dh-elpa experimental->unstable, and rebuild everything, to fix the bug that's
been affecting users upgrading between stable releases.

--
Sean Whitton

--=-=-Content-Type: application/pgp-signature; name="signature.asc"

-----BEGIN PGP SIGNATURE-----

iQJNBAEBCgA3FiEEm5FwB64DDjbk/CSLaVt65L8GYkAFAmagYKMZHHNwd2hpdHRv bkBzcHdoaXR0b24ubmFtZQAKCRBpW3rkvwZiQHGKD/484o+hOm3m5WKVo3LO4973 /oLjPZeRT6+HiLxpqoMgeB1svdwYEprCZ0LCjUXlg9KYofc42tynL6NxRe1QqxL8 +gVxV0MhqnBsW52aQtkT+/1yO1nPpHl7QMxhaZwVuQGRr344GPFobDpDkjjLcE5D bX7nOYmOSHIkXQOBs35Bkt5RJVj1qxv1h5Z1Yn99IMwPaJ0jeyQKZ9VS8tnacL9m y/L3eKoXjSYXAoUPaRCxNn8vjKgFdBPDryT7+zOb2Fu/KhxglBPc1IytQm5sDwPx zNlNZHb1o7i+cZZTC9g4ARoml97h6wN+WEk4C7HCvZv1Ro0JzulLCam2vgmmoPH+ gaJgGp7pG1l87LTblqW567B9Nw0D+O6gzH/Twz7wG69rMdybW6Fc9M78laDkC7Yh 3gvbqJTRtbfNd/0r4nKDpNLudCDCbeWHPhOYmSUPKXdqb8rmzPtniitCrDVBUsRV 3k7chg5GwjPnFVrCuNMvg+Ptx9HsfhJCN6gHzRdXS6vRH2mOiV44iJUNIveX2VkE r2oZNSmxEg+QTDi1Mjv7FHF7ALmOt2F4EAkttD0H2XDN7CYzVEZkQzJtQKzL/1PJ WXfwtNlwWzHVvHyh2i3lTmW/jcsw/TSucebNlsJRUByhtt9fX9qwPJZWOhtsJxQZ 0WF3tjMPqeKFZaW6Apfc/Q==Qgqc
-----END PGP SIGNATURE-----

--- SoupGate-Win32 v1.05
* Origin: fsxNet Us

Rob Browning <rlb@defaultvalue.org> writes:

Sounds reasonable -- any chance y'all are going to have someone who
might know handy at debconf?

Oh, and if you do happen to chat with anyone dpkg-related, as an
alternative, it might be handy if we had a tool (dpkg or other) that
could properly sort a list of packages in dependency order (i.e. to
replace our possibly sketchy tsorting).

(...if we've understood it correctly, the very limited guarantees wrt
the state of dependencies during maintainer script invocations is the
primary complexity. I wonder how many maintainers realize just how
limited they are, if you want to get it *right*.)

--
Rob Browning
rlb @defaultvalue.org and @debian.org
GPG as of 2011-07-10 E6A9 DA3C C9FD 1FF8 C676 D2C4 C0F0 39E9 ED1B 597A
GPG as of 2002-11-03 14DD 432F AE39 534D B592 F9A0 25C8 D377 8C7E 73A4

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Sean Whitton <spwhitton@spwhitton.name> writes:

Thanks. When David and I talked about it, we thought that we shouldn't
rely on ordering that's not guaranteed by the triggers spec -- though possibly the spec has fallen out-of-date with the implementation.

Sounds reasonable -- any chance y'all are going to have someone who
might know handy at debconf?

Thanks
--
Rob Browning
rlb @defaultvalue.org and @debian.org
GPG as of 2011-07-10 E6A9 DA3C C9FD 1FF8 C676 D2C4 C0F0 39E9 ED1B 597A
GPG as of 2002-11-03 14DD 432F AE39 534D B592 F9A0 25C8 D377 8C7E 73A4

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)