[gentoo-dev] Re: GLEP 42 (Critical News Reporting) round five

[Duncan]

Ciaran McCreesh posted <20051213032043.55a6e40f@××××××××.home>, excerpted
below,  on Tue, 13 Dec 2005 03:20:43 +0000:

> Ok, new draft. Changes are as follows:
[]
> * Changed /var/lib/portage to /var/lib/gentoo

OK, I must have missed the reason for that, and it isn't listed in one of
your "a previous version" notes, unless I missed that too. <g>

Assuming the reason wasn't contrary to this (which it probably is,
but...), why not /var/lib/portage/news/gentoo?  If I read jstubbs'
suggestion correctly, the "gentoo" would then serve as the repo name (in
place of magic-chicken, altho as he proposed it, that would be part of the
filename, not the directory the file is in) as well -- he said naming the
current/default repo "gentoo" was sufficient.

> * Added emerge --ask thingie

> Checks for new news messages should be displayed:
[]
> * After an ``emerge --pretend``
[]
> * Before an ``emerge --ask <target>`` sequence

Wouldn't it be less confusing if the news warning appeared in the same
place, relative to the package listing, in both of these?  Isn't an emerge
--ask just the output of pretend, with a confirmation pinned to the end? 
Shouldn't it continue to be that, at least in concept?

> * news.read is now mandatory for interactive clients, and ignored for
> gateway clients

> When a news item is read, its name should be removed from the
> ``news-magic-chicken.unread`` file. If a news client acts as an
> interactive reader rather than a gateway, it should then add the name to
> a ``news-magic-chicken.read`` file in the same directory with the same
> file format (again, ``magic-chicken`` should be a wildcard rather than
> hardcoded).

First, the change outline doesn't state what the result actually was, in
the GLEP. Mandatory would require a MUST (or a similar statement that it's
mandatory), while the GLEP words it as a SHOULD.  Or is "should" not to be
taken in the usual RFC meaning, but rather as an RFC "MUST"?

Second but related, the first time I read thru it, I somehow missed the
"rather than a gateway" part.  Upon rereading, I saw it (obviously), but
the effect of the present wording is to deemphasize the "gateway" clause,
as well as the "read" file.  If it's truly a MUST, then the "read" file
deserves equal treatment with the "unread" file, probably by introducing
the two as a pair, then treating them in parallel thru most of the other
references.

(IOW, the read file and its requirement for interactive clients currently
appears to be the afterthought it in fact was, without that fact
being recognized, which doesn't particularly positively impress,
quality-wise.)

Third, recall from the discussion of an earlier draft, someone mentioned
the multiple meaning of read (as here) vs. "read" (as in README).  The
suggestion to avoid that ambiguity was "seen" and "unseen".  Another might
be (un)viewed.  I'm not sure this is a big enough issue to matter much,
particularly with "unread" there as well, to influence the context, but as
I don't recall that point being addressed, I thought I'd mention it here.

> Read the whole thing before commenting please.

I did.

FWIW & IMO...  Your tenacity and attention to detail are both extremely
good qualities to have in someone doing a GLEP.  Few have the attention to
detail and self-standards necessary, and I fear many that do would give up
due to the barrage of criticism (hopefully all constructive <g>) these
things get.  Do keep up the good work!  IMO, you are /far/ better at it
than most would be, and the resulting GLEP will ultimately be the better
for it!

-- 
Duncan - List replies preferred.   No HTML msgs.
"Every nonfree program has a lord, a master --
and if you use the program, he is your master."  Richard Stallman in
http://www.linuxdevcenter.com/pub/a/linux/2004/12/22/rms_interview.html


-- 
gentoo-dev@g.o mailing list