On Fri, Jan 5, 2018 at 5:08 AM, Ulrich Mueller <u...@gentoo.org> wrote:

> >>>>> On Thu, 4 Jan 2018, Alec Warner wrote:
> > Brief amendment. In the case where the PM cannot parse the expires
> header; it
> > should assume the item is not expired and display it (e.g. it should fail
> > open.)
> > Updated patch attached.
> > +  ``Expires:``
> > +    Date of expiration, in ``yyyy-mm-dd`` format (e.g. 2005-12-18) for
> This is only an example, but choosing a date from 2005 looks strange.
> > +    compatability with GLEP 45 [#glep-45]_. Translations should use the
> date of
> > +    the original news item. An item is expired if the current date in
> UTC is
> > +    greater than the expiration date of the item. Package manages
> should not
> > +    display expired items. In the event where the Expires: header not
> readily
> > +    converted to a date, package managers should assume items are
> unexpired.
> I would strike that last sentence. The GLEP already says "tools
> handling these news items must ignore any unrecognised header" which
> implicitly covers it.

I wrote a lot more because I also wrote the patches to portage and things
were unclear to me.
For example when I read the spec "unrecognized header" to me meant the
header name only;
I'll add some extra words to clarify that invalid values are the same as
invalid header names.

> Also, if we would want to specify more explicitly how to deal with
> invalid header syntax, then there should be a general section or
> paragraph about that.

Ack, I've modified the unrecognised header section to clarify this a bit.

> > +    In news item format ``>2.0``, this field is mandatory.
> I think it should not be mandatory, for the purpose of the tools
> dealing with news items. So I'd simply say here: "Only in news item
> format 2.1 or later."

My concern is that if it isn't mandatory; we will end up with the same

No one will set expiry headers on their items and we will be forced to go
back and update old news items to add them once
there are too many items (today's state.)

Alternatively we could simply state amend the glep to have a default expiry
(say 3y) and if no expires header is present we consume the Posted header
for this purpose.

> >                                                             This field
> did not
> > +    exist in formats ``<=2.0`` and is optional there.
> Strike this sentence. If we say "only in format 2.1" above, then it is
> clear that it didn't exist before.
> In addition, in the paragraphs for the "Display-If-*" headers, the 2.0
> need to be updated to something like "2.0 or later" or "2.*".

Ack, done.

> Ulrich

Reply via email to