DataMuseum.dk

top - metrics - download

    Length: 5266 (0x1492)
    Types: TextFile
    Names: »=RN-5.0«

└─⟦a05ed705a⟧ Bits:30007078 DKUUG GNU 2/12/89
    └─⟦32c6b6024⟧ »./gnews-2.0-tar.Z« 
        └─⟦5d06bd818⟧ 
            └─⟦this⟧ »gnews-2.0/=RN-5.0« 

TextFile

Newsgroups: news.software.b
From: lwall@devvax.JPL.NASA.GOV (Larry Wall)
Subject: Re: Looking for Version 4.4 of rn
Date: 19 Jan 88 07:17:50 GMT
Reply-To: lwall@jpl-devvax.JPL.NASA.GOV (Larry Wall)
Organization: Jet Propulsion Laboratory, Pasadena, CA.

In article <417@n8emr.UUCP> lwv@n8emr.UUCP (Larry W. Virden) writes:
* Anyone know when version 4.4 of rn is slated for release?  I have tried
* to find a path to Larry Wall, but my mail just bounces back to me...

My current address is lwall@jpl-devvax.jpl.nasa.gov.

I have even less idea when 4.4 will come out than you do.  

No, I take that back.  I know exactly when 4.4 will come out.

Never.    :-)

Don't panic...it's just that there are so many changes that it ought
to be called 5.0 when it comes out.  This is a major rewrite, folks.
There have been some major changes in the internal organization of rn.
The main one is called "virtual articles".  The display routines used
to directly call file access routines.  Now they can call an entire
pipeline of routines, specifiable on a newsgroup by newsgroup basis,
that let you say where to get the articles from (spool dir, NNTP, mbox,
personal archives, etc...) and how to format the article (highlighting,
auto rot13, header line suppression, etc.).

The other major change is that you will be able to deal with sets of
articles as named entities.  These sets can be defined by rule or
enumeration.  Sets defined by rule can have new articles added to them
automatically.  You can define set generators that group articles into
sets by, say, subject, or anything else you can specify via a % string.

Kill processing and lookahead functions will work differently.  Kill
processing falls out as a special case of preprocessing articles into
article sets--they just happen to fall into sets you don't want to see.

The .newsrc will be scrapped in favor of a .rnrc, which will contain
much more info.  A .newsrc will be constructable from a .rnrc, and vice
versa.  An entry in .rnrc will look something like this:

comp.foo.bar: 1-12345,12347-12349,12351
(SF 12345)
(TH 1234)
(IN =)
(SW -ESUBJLIST="%S %D")
(PI nntp | undig | autorot || myhead | hide | hilite)
(SV ~/amiga/%s)
(HD /^>/)
(HI /\brn\b/)
(KL
	53:/foo/a:j
)
(CN
	%s %s
)
(ST
	321:/transmogrify/
	/bar/ 1,3,5,7
	/blah blah/ 78-82
	/My machine is better than yours/ 38,42-43,92
)
(DG
	123=5 513,1020,2192,3283,4182,5129
)

Most newsgroups will not have all these fields.  Note that the first line
of each entry is the classical .newsrc line.  You can make the corresponding
.newsrc with grep.  You'll be able to generate a .rnrc from a .newsrc by
simply appending the .newsrc to the end of the .rnrc that you last used.

Explanation of the fields
(SF 12345)
	A soft pointer into the active file.

(TH 1234)
	The THRU counter that used to be at the front of the KILL file.

(IN =)
	Any commands to do upon entering the newsgroup.

(SW -ESUBJLIST="%S %D")
	Any switches to set upon entering the newsgroup.  They are
	automatically restored to their former value upon exiting the
	newsgroup.

(PI nntp | undig | autorot | myhead | hide | hilite)
	The source and formatting of the articles in this newsgroup.
	This line says take articles via NNTP, undigestify them, run the
	autorotator on them, run through a heading customizer (that uses
	% interpolations to build the header any way you choose), run
	through a routine that hides unwanted lines, and a routine that
	highlights any specifiable patterns.  All these filters will be
	built in, but I expect if I don't recognize a filter name I'll
	just pass it off to the shell, if you're willing to take the
	overhead.

(SV ~/amiga/%s)
	Default location to save files from this newsgroup.

(HD /^>/)
	Pattern of lines that the hide filter hides.

(HI /\brn\b/)
	Patterns that the hilite filter highlights.

(KL
	53:/foo/a:j
)
	KILL set definitions.

(CN
	%s %s
)
	Set constructors--this one says classify by subject, use subject
	for the set name.

(ST
	321:/transmogrify/
	/bar/ 1,3,5,7
	/blah blah/ 78-82
	/My machine is better than yours/ 38,42-43,92
)
	Current sets.  They can contain articles that are marked read, but
	the list-subjects command can do set intersections with the set of
	unread articles.

(DG
	123=5@1438 513,1020,2192,3283,4182,5129
)
	Undigestification data used by undig filter.  This says something
	like real article 123 turns into 5 articles starting at virtual
	article number 1438, with the following offsets into the real
	article.

All this is subject to change, of course.  And there are other ideas
floating around, such as if you run "rn -c" in your .login file it starts
up a background rn that starts crunching through all your preprocessing
for you until you run rn for real.  Or how about n-level undo?

There's no limit on the number of newsgroups any more.

There's a new newsgroup state "To Be Added" that will let you add new
newsgroups at your leisure rather than forcing you to do it all now.

Now for the bad news.  I don't think it'll run on a PDP-11.

All this is about half done.  Which means things are very messy.  I drag it
out and work on it once every few months.  Perhaps now that perl is coming
out, I can look at it more often.  But call it 5.0.

Larry Wall
lwall@jpl-devvax.jpl.nasa.gov