[Qemu-devel] qemu-devel archive search was missing some messages but it's fixed now

[Qemu-devel] qemu-devel archive search was missing some messages but it's fixed now

[Ian Kelling]

Hi. I work at FSF and help maintain the mailing lists software.

I don't know exactly how many were missing because there was an error we
were ignoring. At least some from November. I fixed it by reindexing
from scratch. I didn't figure out the the root cause, likely something
to do with incremental indexing. But we are no longer ignoring errors,
so we will fix issues if they occur again.

General thoughts about the archive: I'm looking forward to us upgrading
to mailman 3 at some point which will bring an improved archive
interface. Of course, you can always download the archive and search it
on your local computer.

-- 
Ian Kelling | Senior Systems Administrator, Free Software Foundation
GPG Key: B125 F60B 7B28 7FF6 A2B7  DF8F 170A F0E2 9542 95DF
https://fsf.org | https://gnu.org

Re: [Qemu-devel] qemu-devel archive search was missing some messages but it's fixed now

[Laszlo Ersek]

Hello Ian,

On 12/04/17 21:02, Ian Kelling wrote:

> General thoughts about the archive: I'm looking forward to us upgrading
> to mailman 3 at some point which will bring an improved archive
> interface.

Do you mean HyperKitty?

Will you seek out community feedback on it first?

> Of course, you can always download the archive and search it
> on your local computer.

The one HyperKitty instance that I can readily look at is:

https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/

It only offers the following download options:

- "Past 30 days (mbox)"
- "This month (mbox)"
- "Entire archive (mbox)"

Maybe it's configurable site by site, but I find this lacking compared
to: "download whichever month you want, in isolation", which is offered
by mailman2:

ftp://lists.gnu.org/qemu-devel/

If someone needs it all, the latter remains easy to script with wget,
for example.

(Back to my original point -- judged similarly from the fedora devel
archive, I find HyperKitty hardly usable for normal reading, and I think
it will completely fail for patch series threads and the discussions
under them. HyperKitty is a step towards social media in my opinion, and
as such inappropriate for development lists (again, IMO). I'm happy to
learn otherwise, of course -- are there good counter-examples where
HyperKitty has worked out fine, for archiving email-based patch traffic
and development discussion?)

NB: my opinion on this likely doesn't matter, as I'm not a heavy
qemu-devel contributor.

Thanks
Laszlo

Re: [Qemu-devel] qemu-devel archive search was missing some messages but it's fixed now

[Ian Kelling]

Laszlo Ersek <lersek@redhat.com> writes:

> Hello Ian,
>
> On 12/04/17 21:02, Ian Kelling wrote:
>
>> General thoughts about the archive: I'm looking forward to us upgrading
>> to mailman 3 at some point which will bring an improved archive
>> interface.
>
> Do you mean HyperKitty?

Yes.

>
> Will you seek out community feedback on it first?

Yes, of course. And I expect the old interface to continue to exist
alongside it indefinitely.

>
>> Of course, you can always download the archive and search it
>> on your local computer.
>
> The one HyperKitty instance that I can readily look at is:
>
> https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org/
>
> It only offers the following download options:
>
> - "Past 30 days (mbox)"
> - "This month (mbox)"
> - "Entire archive (mbox)"
>
> Maybe it's configurable site by site, but I find this lacking compared
> to: "download whichever month you want, in isolation", which is offered
> by mailman2:
>
> ftp://lists.gnu.org/qemu-devel/
>
> If someone needs it all, the latter remains easy to script with wget,
> for example.
>
> (Back to my original point -- judged similarly from the fedora devel
> archive, I find HyperKitty hardly usable for normal reading, and I think
> it will completely fail for patch series threads and the discussions
> under them. HyperKitty is a step towards social media in my opinion, and
> as such inappropriate for development lists (again, IMO). I'm happy to
> learn otherwise, of course -- are there good counter-examples where
> HyperKitty has worked out fine, for archiving email-based patch traffic
> and development discussion?)
>
> NB: my opinion on this likely doesn't matter, as I'm not a heavy
> qemu-devel contributor.

And you've given some reasons why :)

-- 
Ian Kelling | Senior Systems Administrator, Free Software Foundation
GPG Key: B125 F60B 7B28 7FF6 A2B7  DF8F 170A F0E2 9542 95DF
https://fsf.org | https://gnu.org

Re: [Qemu-devel] qemu-devel archive search was missing some messages but it's fixed now

[Ian Kelling]

Related, I've updated
https://lists.nongnu.org/archive/cgi-bin/namazu.cgi?idxname=qemu-devel
with new text to explain what is configured to not be in the search index:

Long messages and words are ignored

Messages longer than 100,000 letters or 500,000 bytes are ignored. Words
longer than 40 characters are ignored. Attachments are ignored.

-- 
Ian Kelling | Senior Systems Administrator, Free Software Foundation
GPG Key: B125 F60B 7B28 7FF6 A2B7  DF8F 170A F0E2 9542 95DF
https://fsf.org | https://gnu.org