[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Less dinstall FTW?

Ansgar Burchardt writes ("Re: Less dinstall FTW?"):
> as I have seen some confusion what this change means in practice and
> most answers ignored the second part of the proposal, here are some more
> explanations:

Thank you for the clear explanation.  I'm much less confused.

> The latter is probably the main reason people are interested in
> dinstall: uploaded packages get accessible. If your work is blocked, but
> a fix was recently uploaded, you currently have to either wait for
> dinstall and your mirror to be pushed or search and download the file in
> question by hand on incoming.d.o (and find the signature in the mail
> archive). This is one of the reasons some people want to have dinstall
> run more often.


> The more interesting part of the proposal has so far been ignored by
> most replies: we would make the incoming.d.o archive public. This would
> mean all new uploads are available after ~15 minutes via APT, a lot
> faster than the current interval between dinstall runs.


My concern in this conversation is to maximise the
automatically-findable availability of packages which the dak database
considers to be published.

I like to look at the archive as a VCS.  Looking at it like that, for
a long time it has been something of a deficiency that it can be
difficult to find out what version of something is in the archive, and
obtain that version.

In particular, in the context of the current discussion, I would like
to be able to do "dak ls" and then be able to somehow actually
retrieve the corresponding files - automatically.  (Given that our
filenames are not reused, it is fine to try a mirror first, but I want
to be able to fall back to something close to the source.)

It sounds like these proposals will make this easier.


Reply to: