Newsgroups: why limit getting headers to 251?
Boris Borisovich
all I can get in 251 headers. This is very silly - how do I default to get
all headers?
As a result, I have had to search for this answer in xnews, rather than
Opera.
thanks
Boris
Richard Grevers
wrote:
> I'm just giving Opera a spin - on trying to get news from my newsserver,
> all I can get in 251 headers. This is very silly - how do I default to
> get
> all headers?
>
250 is a hard-coded limit, compared to some news clients which let you set
the answer.
Once you have the 250 headers, chec the message number of the first one,
then use an url such as
news://news.opera.com/opera.general/1-72861
where you would make the second number one less than the first header you
have.
Note however, that on a typical ISP newsfeed, older headers are expired to
save space and typically there will be fewer than 250 headers available.
--
Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/
Boris Borisovich
news:op.ssy6jpcvigav4j@richardg:
thanks Richard for the explanation, but I think Opera's default is
ridiculous: my newsserver goes back months, and I find searching for
answers to things very easy in xnews, which gets everything by default.
Almost always someone has asked the very quetion I want answered, and I
don't have to annoy people by reposting the same question twice. Getting
just 251 headers gives me a weekend's worth of headers.
I thought Opera was good - this is first chink in the armour.
thanks
Boris
FACE
In many groups that would give you a few hours worth.
I just sampled the 5000 most recent headers from uk.politics.misc and at
that time I got the message "retrieving 5000 of 168,719 headers"
250 is ridiculous.
FACE
Boris Borisovich
news:kr6tb11vpjsq129kt...@4ax.com:
snip
> In many groups that would give you a few hours worth.
>
> I just sampled the 5000 most recent headers from uk.politics.misc and at
> that time I got the message "retrieving 5000 of 168,719 headers"
>
> 250 is ridiculous.
>
> FACE
>
FACE, we are in violent agreement
Boris
FV
13:36:51 +0200:
> 250 is a hard-coded limit, compared to some news clients which let you
> set the answer.
> Once you have the 250 headers, chec the message number of the first one,
> then use an url such as
>
> news://news.opera.com/opera.general/1-72861
> where you would make the second number one less than the first header
> you have.
Note that threading in Opera will fail if a previous message in a thread
comes in later. This is a regression between 7.2 and 7.5 that has never
been fixed, AFAIK.
--
Fabian
FV
+0200:
> thanks Richard for the explanation, but I think Opera's default is
> ridiculous: my newsserver goes back months, and I find searching for
> answers to things very easy in xnews, which gets everything by default.
> Almost always someone has asked the very quetion I want answered, and I
> don't have to annoy people by reposting the same question twice. Getting
> just 251 headers gives me a weekend's worth of headers.
>
> I thought Opera was good - this is first chink in the armour.
Indeed any default here is ridiculous; the user should simply have the
choice. The same goes for Opera's downloading every message in an already
subscribed group.
I make very limited use of the newsreader and can live with it, but I can
understand there are many people who simply need more options. Opera's
news client has been very basic from the start. Improving it is on the
agenda, but the developers seem to have had other priorities continuously,
which causes the current newsreader to have a state not much better (in
some aspects worse) than Opera 7.0, 2½ years ago.
If Opera wants to have a decent newsreader built-in, perhaps it would be
an idea to hire a programmer who can focus on this task.
--
Fabian
Boris Borisovich
news:op.ssy9x...@rainbow.wanadoo.nl:
snip
> Note that threading in Opera will fail if a previous message in a
> thread comes in later. This is a regression between 7.2 and 7.5 that
> has never been fixed, AFAIK.
>
this is very disappointing - I don't think Opera can sell itself as the
all-singing all-dancing internet app. if it can't handle NG's. OTOH, NG's
are rather going out of fashion in favour of web-based forums, so perhaps
it is bottom of the pile as far as Opera's development goes.
Boris
Gene Goldenfeld
While I don't use Opera mail/newsreader, preferring instead NS 4.8
Messenger with its limitations, the point that ISPs and newservices vary
widely on how far back they go is well taken. My current one seems to
go back 24 months, but others have been as short as two or three
months. If I want to do a search on content, I find google groups the
handiest.
Gene
Gene
Richard Grevers
<gene...@highstream.net> wrote:
>
> While I don't use Opera mail/newsreader, preferring instead NS 4.8
> Messenger with its limitations, the point that ISPs and newservices vary
> widely on how far back they go is well taken. My current one seems to
> go back 24 months, but others have been as short as two or three
> months. If I want to do a search on content, I find google groups the
> handiest.
>
All NZ ISP's get their feeds from two main providers, and they are both:
nz.* 4 weeks, big 7, 1 week, alt.* 72 hours
Steven V. Gunhouse
wrote:
When you first subscibe, Opera defaults to 250. After that it gets all new
messages.
There are two strategies to fetch all (available) headers from a group.
The one I like best is to edit the newsrc file for that server, someone
recently reported complications with the alternative method...
If you look in your mail folder, you'll find a file named
(servername).newsrc which contains the names of all the newsgroups
available on the server and your current status with respect to each group
(subscribed or not and last message read). Here, let me quote part of mine
(this is my ISP's server, so it contains a number of non-Opera groups too):
okinawa.mail-lists.nirai-kanai!
opera.general: 0-69728
powersoft.public.powerbuilder.distributed!
prg.jobs!
The ":" indicates that I'm subscribed to opera.general, and the number
69728 is the last message Opera has seen; the "!" is groups which I
haven't sobscribed to.
If I were to change the 69728 to 0, then the next time I logged in Opera
would download however many messages that my server has for that group.
Given that it is my ISP's server, maybe 5000-10000. If you subscribed to
Opera's server directly, then that could be several thousand - probably
not all 69728, some messages were cancelled by the original sender and I'd
hope they've removed the few spam messages that get sent to the list, but
even 50000 might be more than you really want ...
--
Using Opera's revolutionary e-mail client: http://www.opera.com/m2/
djk
bo...@home.com says...
> this is very disappointing - I don't think Opera can sell itself as the
> all-singing all-dancing internet app. if it can't handle NG's. OTOH, NG's
> are rather going out of fashion in favour of web-based forums, so perhaps
> it is bottom of the pile as far as Opera's development goes.
>
>
I don't use Opera for mail and I don't use Opera for
newsreading. It's unfortunate that Opera sells itself as ready
for prime-time for mail & news when it is not but I don't really
care about that marketing decision because as a browser it is
worth what I paid.
Matthew Winn
>
> this is very disappointing - I don't think Opera can sell itself as the
> all-singing all-dancing internet app. if it can't handle NG's. OTOH, NG's
> are rather going out of fashion in favour of web-based forums, so perhaps
> it is bottom of the pile as far as Opera's development goes.
Outside the US (where "the web" and "the internet" are seen as
synonymous) I don't think newsgroups are going out of fashion.
Web-based discussion areas are so primitive compared with usenet
that I doubt they'll ever completely take over as a serious forum,
though they certainly do appeal to the sort of people who think
that it's just not possible to make a point without having a dozen
winking smilies after each sentence.
But as for Opera's support for newsgroups, it is indeed rather
disappointing. How difficult can it be to add a UI field for
specifying the number of entries to fetch from a new group?
--
Matthew Winn
[If replying by email remove the "r" from "urk"]
Boris Borisovich
<svgun...@ameritech.net> wrote:
snip
thanks for that - it seems to work on this first attempt, but it is an
incredibly clunky way of doing something so basic - how could any
developer possibly come to the view that all any user wanted was 250
headers? One of hte really great things about NG's like this one is that a
good server will have information going back years, so making searching
for answers to problems like this one so easy.
thanks for the help
Boris
--
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/
David W. Hodgins
> I'm just giving Opera a spin - on trying to get news from my newsserver,
> all I can get in 251 headers. This is very silly - how do I default to get
> all headers?
See http://groups-beta.google.com/group/opera.mail+news/msg/1fb64c61ad16fa00?dmode=source&hl=en
Regards, Dave Hodgins
--
Change nomail.afraid.org to rogers.com to reply by email.
(nomail.afraid.org has been set up specifically for
use in usenet. Feel free to use it yourself.)
Richard Grevers
wrote:
>
> thanks for that - it seems to work on this first attempt, but it is an
> incredibly clunky way of doing something so basic - how could any
> developer possibly come to the view that all any user wanted was 250
> headers? One of hte really great things about NG's like this one is that
> a good server will have information going back years, so making
> searching for answers to problems like this one so easy.
>
Um, because downloading and storing thousands of posts (several MB) from a
group in order to search them is quite a bizarrely inefficient solution to
a problem compared with searching the usenet archive (google groups).
--
Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/
Rijk van Geijtenbeek
wrote:
..
> thanks for that - it seems to work on this first attempt, but it is an
> incredibly clunky way of doing something so basic - how could any
> developer possibly come to the view that all any user wanted was 250
> headers? One of hte really great things about NG's like this one is that
> a good server will have information going back years, so making
> searching for answers to problems like this one so easy.
Getting all messages from the start of the newsgroups on this server will
take hours, depending on your connection speed of course. In the absence
of configurable options, 250 is a good default. Of course, everyone wants
configurable options, or a choice per group. No dispute there.
--
The Web is a procrastination apparatus: | Rijk van Geijtenbeek
It can absorb as much time as | Documentation & QA
is required to ensure that you | Opera Software ASA
won't get any real work done. - J.Nielsen
|http://my.opera.com/Rijk/journal
Peter Boulding
wrote in <op.ss05f...@laptop-rock.lan>:
>Getting all messages from the start of the newsgroups on this server will
>take hours, depending on your connection speed of course. In the absence
>of configurable options, 250 is a good default. Of course, everyone wants
>configurable options, or a choice per group. No dispute there.
Given that broadband users can download messages between five and twenty
times faster than modem users, there is no "good default."
A configurable option in this respect is one of the sine qua nons of a
usable news client. This is one of a number of failings I have learned of in
this newsgroup that make me wonder why anyone other than an unthinkingly
fanatical Opera lover would consider using this news client.
--
Regards
Peter Boulding
p...@UNSPAMpboulding.co.uk (to e-mail, remove "UNSPAM")
Fractal music & images: http://www.pboulding.co.uk/
Rijk van Geijtenbeek
<p...@UNSPAMpboulding.co.uk> wrote:
..
> Given that broadband users can download messages between five and twenty
> times faster than modem users, there is no "good default."
>
> A configurable option in this respect is one of the sine qua nons of a
> usable news client.
Depends on your uses for newsgroups.
> This is one of a number of failings I have learned of in this newsgroup
> that make me wonder why anyone other than an unthinkingly
> fanatical Opera lover would consider using this news client.
If you have so much time you want go through thousands of old messages, or
if you want to use the binaries groups, Opera is not yet the best
newsgroups reader. There are lots of people using this newsgroups client
without problems. But it will not be a special reason to start using
Opera, IMHO. I don't think you'll find the Opera fans here telling you
otherwise.
Lauri Raittila
> If you have so much time you want go through thousands of old messages [snip]
> Opera is not yet the best newsgroups reader.
Are you sure? Actually I think its ability on that respect is far
superior to FreeAgent, Xnews, Gravity, Pine and propably many others...
Because it can actually search from those thousands of old messages, and
it wont take ages. Or that was my impression when I tried last time, long
time ago...
> There are lots of people using this newsgroups client
> without problems. But it will not be a special reason to start using
> Opera, IMHO. I don't think you'll find the Opera fans here telling you
> otherwise.
Yes. Opera combined browser and mail suite, with some rss, irc, ftp, and
usenet additions to do the very basics.
--
Lauri Raittila <http://www.iki.fi/lr> <http://www.iki.fi/zwak/fonts>
Kohtuuhintainen yksiö/huone haussa Oulusta syyskuusta eteenpäin.
Searching places to sleep on axis Bonn - Tsech - Poland - baltic sea in
july
Rijk van Geijtenbeek
> in opera.general, Rijk van Geijtenbeek wrote:
>
>> If you have so much time you want go through thousands of old messages
>> [snip]
>> Opera is not yet the best newsgroups reader.
>
> Are you sure? Actually I think its ability on that respect is far
> superior to FreeAgent, Xnews, Gravity, Pine and propably many others...
Yes... if you have built up a collection of received messages already. If
you would subscibe to a new group to find a specific answer, tapping the
collective wisdom from Google Groups is usually a faster option.
> Because it can actually search from those thousands of old messages, and
> it wont take ages. Or that was my impression when I tried last time, long
> time ago...
>
>> There are lots of people using this newsgroups client
>> without problems. But it will not be a special reason to start using
>> Opera, IMHO. I don't think you'll find the Opera fans here telling you
>> otherwise.
>
> Yes. Opera combined browser and mail suite, with some rss, irc, ftp, and
> usenet additions to do the very basics.
>
--
Get Opera 8 now! Speed, Security and Simplicity.
http://my.opera.com/Rijk/affiliate/
Rijk van Geijtenbeek
Opera Software ASA, Documentation & QA
Tweak: http://my.opera.com/Rijk/journal
Lauri Raittila
> On Tue, 28 Jun 2005 01:16:08 +0200, Lauri Raittila wrote:
>
> > in opera.general, Rijk van Geijtenbeek wrote:
> >
> >> If you have so much time you want go through thousands of old messages
> >> [snip]
> >> Opera is not yet the best newsgroups reader.
> >
> > Are you sure? Actually I think its ability on that respect is far
> > superior to FreeAgent, Xnews, Gravity, Pine and propably many others...
>
> Yes... if you have built up a collection of received messages already. If
> you would subscibe to a new group to find a specific answer, tapping the
> collective wisdom from Google Groups is usually a faster option.
But google groups won't have that good search. It is not much problem
when stuff is in English or Dutch, but in Finnish groups it is sometimes
really annoying to try guess which form of the word was used. (10 most
used is good bet, but that already needs 10 terms for google.
As local storage space for messages is no longer probelm, as it was few
years ago, using your own archve might be actually quite handy.
Some Finnish servers also have up to 3 years of messages, so it is not
that far fetched.
Downloading and organizing newsgroup messages takes maybe 2 minutes using
Opera, counting only time needed for action. As download time is far
greater, I think this is the place where Opera is better than other
newsclients.
Richard Grevers
<la...@raittila.cjb.net> wrote:
> in opera.general, Rijk van Geijtenbeek wrote:
>> On Tue, 28 Jun 2005 01:16:08 +0200, Lauri Raittila wrote:
>>
>> > in opera.general, Rijk van Geijtenbeek wrote:
>> >
>> >> If you have so much time you want go through thousands of old
>> messages
>> >> [snip]
>> >> Opera is not yet the best newsgroups reader.
>> >
>> > Are you sure? Actually I think its ability on that respect is far
>> > superior to FreeAgent, Xnews, Gravity, Pine and propably many
>> others...
>>
>> Yes... if you have built up a collection of received messages already.
>> If
>> you would subscibe to a new group to find a specific answer, tapping the
>> collective wisdom from Google Groups is usually a faster option.
>
> But google groups won't have that good search. It is not much problem
> when stuff is in English or Dutch, but in Finnish groups it is sometimes
> really annoying to try guess which form of the word was used. (10 most
> used is good bet, but that already needs 10 terms for google.
But google can also search 10 groups at once.
>
> As local storage space for messages is no longer probelm, as it was few
> years ago, using your own archve might be actually quite handy.
But download volume=money (here, at any rate)
>
> Some Finnish servers also have up to 3 years of messages, so it is not
> that far fetched.
>
> Downloading and organizing newsgroup messages takes maybe 2 minutes using
> Opera, counting only time needed for action.
Opera freezes for 3s (on a 1.6GHz machine) when it pulls in 10 message
bodies in one go (as it indexes them) - how long is indexing 2000 messages
going to take?
--
Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/
Howard Brazee
On 27-Jun-2005, "Rijk van Geijtenbeek" <ri...@opera.com> wrote:
> Getting all messages from the start of the newsgroups on this server will
> take hours, depending on your connection speed of course. In the absence
> of configurable options, 250 is a good default. Of course, everyone wants
> configurable options, or a choice per group. No dispute there.
There may be a good default for a particular broadband speed. Or maybe not.
Certainly someone with a fibre optic connection does not want the same "good
default" as someone with a 300 baud modem.
Howard Brazee
On 27-Jun-2005, "Rijk van Geijtenbeek" <ri...@opera.com> wrote:
> If you have so much time you want go through thousands of old messages, or
> if you want to use the binaries groups, Opera is not yet the best
> newsgroups reader. There are lots of people using this newsgroups client
> without problems. But it will not be a special reason to start using
> Opera, IMHO. I don't think you'll find the Opera fans here telling you
> otherwise.
When they complete M2 to get around its flaws, it has some features that will
make it very attractive. Some of these work better with a full database of
messages to search through, even if we don't go through all of those messages by
hand.
Rijk van Geijtenbeek
wrote:
My point was, that in the *absence* of configurability, 250 is a good
choice. Whatever the speed, downloading all 107370 messages from the start
of this newsgroup is not likely to be the right default choice.
Richard Grevers
wrote:
>
> My point was, that in the *absence* of configurability, 250 is a good
> choice. Whatever the speed, downloading all 107370 messages from the
> start of this newsgroup is not likely to be the right default choice.
>
Does the server polling in the news protocol allow for the client to
contact a server, get the number of messages available and the age of the
oldest message? If so, it would be great when Opera *does* get
configurability if it could report to the user "There are 3,647 messages
available dating back to 7/12/2004. Download latest []"
One snag being that some servers store older headers after they have
discarded the bodies.
--
Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/
djk
la...@raittila.cjb.net says...
> in opera.general, Rijk van Geijtenbeek wrote:
>
> > If you have so much time you want go through thousands of old messages [snip]
> > Opera is not yet the best newsgroups reader.
>
> Are you sure? Actually I think its ability on that respect is far
> superior to FreeAgent, Xnews, Gravity, Pine and propably many others...
>
> Because it can actually search from those thousands of old messages, and
> it wont take ages. Or that was my impression when I tried last time, long
> time ago...
>
>
the database. I've never used Opera as a newsreader but I would
imagine to search bodies of messages it needs to have those
messages in the database. Are you saying Opera searches bodies
on a remote server?
djk
says...
> My point was, that in the *absence* of configurability, 250 is a good
> choice. Whatever the speed, downloading all 107370 messages from the start
> of this newsgroup is not likely to be the right default choice.
>
>
>
in my mind, is not saying a whole lot. I mean, 200 or 300 or
275, basically whatever, would be a good choice with no options.
Torbjorn Lindgren
Well, the reply to the group command gives low and high article
number, and an estimate of the number of articles in the group (IIRC
must not be lower than the actual number of articles, but higher IS
ok, so servers that doesn't know will send HIGH-LOW+1).
The obvious next step is to peek at the overview for early articles,
but as you note that can be unreliable (on "broken" servers). However
I doubt that actually changes matter much, and besides to a large
extent it's "proportions" that matters, and they are likely even
closer...
It wouldn't be that hard to dig out the "oldest" article using
ordinary commands (or even the X oldest, to find the median to dump
bogus dates, next/head seems tailored for this), they aren't affected
by this (just the overview and possibly xhdr).
However, for a server where this will make a significant difference
it's likely that the *estimate* on the number of articles actually
available is also off, and the real count can't be recovered except by
checking all articles. Without this count having removed a few
non-existing articles doesn't help much.
With proper command pipelining it shouldn't take much longer than
sampling overviews on a good server, it does simplify some things it
the calculation but complicates other... It could be slower on "slow"
servers, or if the client doesn't have a command pipelining engine it
might cause some slowdown on high latency links
No, I've not checked if Opera pipelines NNTP commands or not...
Howard Brazee
On 29-Jun-2005, djk <d...@spam.operamail.com> wrote:
> true. but only true in the *absence of configurability*. which,
> in my mind, is not saying a whole lot. I mean, 200 or 300 or
> 275, basically whatever, would be a good choice with no options.
If I could only drive my car at one speed, what speed should it be?
If a ship could only come with one size, what cargo capacity should it have?
If a book could only come in one size, how many words should it have?
If M2 should be changed because it doesn't give us a choice of how many messages
it should download - should the change be in how many messages is hard coded
into the program?
Richard Grevers
No, Opera also needs to store all the message bodies to be searched. But
Opera indexes every word in a message when it arrives, and as a result is
able to run a search on a 100,000 message archive in under 2 seconds,
compared with about a minute for most mail and news applications.
--
Using M2, Opera's revolutionary e-mail client: http://www.opera.com/m2/
John H Meyers
and drop messages older than some specified age, one might
be able to ascertain that if any message in any one group
is now no longer on the server, then it is pretty safe to assume
that all older messages in that group have expired,
and this could be used in a simple way for knowing
when to manually purge expired messages from M2.
However, I can not get my Opera 8.01 (or any earlier version)
to even tell me whether or not any given message remains
on the news server when I try to display it -- all I can get
is "Message body not downloaded" in the message window,
and if I wait and wait and wait and wait some more,
and if nothing further ever happens, then I begin to
*suspect* that perhaps the news server responded that
it didn't still have that message, although Opera
never passes such a response back to me in any visible form.
I should mention that in my news account Properties > Incoming
I have no options checked, which means in particular that
I do not wish to retain newsgroup message bodies at all,
so every time I display a message, I presume that Opera
will be going back to the news server for it
and that Opera ought then to be able to know
when a news server says "I can't retrieve that message"
although I can never really get Opera to tell me that for sure.
All the same, I love the Opera browser, and like the style of M2
well enough to use it for newsgroups, though a few things remain
somewhat unperfected, not quite ready for mission-critical stuff.
Thanks for your interest.
djk
@dramatic.co.nz says...
> No, Opera also needs to store all the message bodies to be searched. But
> Opera indexes every word in a message when it arrives, and as a result is
> able to run a search on a 100,000 message archive in under 2 seconds,
> compared with about a minute for most mail and news applications.
>
>
>
file. Should it become corrupted due to a number of reasons, is
there a way to rebuild it from scratch without an
unsubscribe/subcribe/download messages action?
djk
says...
> If M2 should be changed because it doesn't give us a choice of how many messages
> it should download - should the change be in how many messages is hard coded
> into the program?
>
>
be changed then a *good* choice for the default number could be
any number between 200 & 500 and that choice would be *as good*
as the present choice. There is nothing that *good* about the
current default.
Richard Grevers
Actually it seems to be split among several files. And yes, you simply
delete the lexicon folder and on next startup Opera will commence
reindexing in the background. On a slower machine the indexing can even be
split across several sessions.
djk
@dramatic.co.nz says...
> Actually it seems to be split among several files. And yes, you simply
> delete the lexicon folder and on next startup Opera will commence
> reindexing in the background. On a slower machine the indexing can even be
> split across several sessions.
>
> --
>
>
start again. gravity does a *backup* process that puts all
current settings and database info into a file that can be
restored if necesssary. Does Opera offer this?
Bart Wakker
> thanks for that - it seems to work on this first attempt, but it is an
> incredibly clunky way of doing something so basic - how could any
> developer possibly come to the view that all any user wanted was 250
> headers? One of hte really great things about NG's like this one is
> that a good server will have information going back years, so making
> searching for answers to problems like this one so easy.
You don't seem to have grasped the nature of usenet (newsgroups). It
is not a static archive of information/discussions (except for google
groups of course). Newsservers expire articles after some time (1 day
to maybe a few months). Usenet is intended for actual discussions (my
answer to your post 4 days later is already reaching the limit), so
usually you only need a few hundred headers when first reading a
group. The NNTP protocol is not intended for large scale searches.
If you want to search usenet history, use groups.google.com
djk
> You don't seem to have grasped the nature of usenet (newsgroups). It
> is not a static archive of information/discussions (except for google
> groups of course). Newsservers expire articles after some time (1 day
> to maybe a few months). Usenet is intended for actual discussions (my
> answer to your post 4 days later is already reaching the limit), so
> usually you only need a few hundred headers when first reading a
> group. The NNTP protocol is not intended for large scale searches.
>
> If you want to search usenet history, use groups.google.com
>
>
server. what you don't understand is that by using google you
get only the messages that have not up to this point been
deleted for one reason or another from the google news groups
server. google is more like a web based forum than a true Usenet
archive. That is the fact of google. To get a true rendering of
any Usenet newsgroup you must have access to a good Usenet
newsgroup server and use a newsreader client. And in conclusion,
in reality, and with a good news server, the NNTP protocol *IS*
intended for large scale searches. You just have not been around
long enough to realize this.
Richard Grevers
You are presumable talking about messages posted with a x-no-archive
header. If an nntp server holds on to these, then the nntp server is
broken.
Boris Borisovich
server.cfl.rr.com:
snip
> problem is you are a recent disciple of google newsgroups
> server. what you don't understand is that by using google you
> get only the messages that have not up to this point been
> deleted for one reason or another from the google news groups
> server. google is more like a web based forum than a true Usenet
> archive. That is the fact of google. To get a true rendering of
> any Usenet newsgroup you must have access to a good Usenet
> newsgroup server and use a newsreader client. And in conclusion,
> in reality, and with a good news server, the NNTP protocol *IS*
> intended for large scale searches. You just have not been around
> long enough to realize this.
good point - I think this is a bad point in opera, it should mend it
SOON,or I won't pay for it.
Boris
djk
@dramatic.co.nz says...
> You are presumable talking about messages posted with a x-no-archive
> header. If an nntp server holds on to these, then the nntp server is
> broken.
>
>
the result of the use of a good Usenet News server *and* a news
client. I forgot to mention that the client should download both
headers and bodies, but i do so at this time. You are correct
that a nntp server *should* not archive "x-no-archive" posts but
someone that had access to the server to download any such posts
would have them archived. Big point is that google is *not* a
source of a true & complete archive. There are still servers out
there that do have a fairly deep (aged) archive. Historically
nntp servers have had excellent archives, I'm talking 1985
through as recent as 2003. The problem is that Usenet is being
attacked by posters who flood Usenet with useless posts thus
creating a volume problem which affects retention times of
useful posts. Main issue was the claim that nntp was no meant
for archive. It was and it worked wonderfully until Usenet
became more mainstream and filled with the amount of nonsense
that it is today.
Torbjorn Lindgren
>Since news servers generally receive the spool in time sequence,
>and drop messages older than some specified age, one might
>be able to ascertain that if any message in any one group
>is now no longer on the server, then it is pretty safe to assume
>that all older messages in that group have expired,
>and this could be used in a simple way for knowing
>when to manually purge expired messages from M2.
There are many reasons why this might not always be true (and I've
probably forgotten some reasons):
* Cancelled messages (though this is often disabled)
* Messages with explicit expiration dates (this is honored more often)
* Some servers use different expiration dates for different "types" of
messages, even in one group (single article messages saved for months
or even years, multipart articles expired pretty soon, days to weeks).
On the other hand, IF all earlier messages are indeed deleted the low
watermark from the group command should have been adjusted, and in
practice it is almost always keept very close to the real value, so
there should be no need for this crude and fallible heurestic.
The client ought to send group commands now and then to update both
low and high watermark anyway, so that it knows what's happening with
the group and can grab new headers...
The most likely cause for large swaths for "unavailable" articles
inside the "active" area is there are indeed one or more articles with
low article numbers left, that the client isn't updating the low
watermark or a a very broken server, get a new one... :-).
>However, I can not get my Opera 8.01 (or any earlier version)
>to even tell me whether or not any given message remains
>on the news server when I try to display it -- all I can get
>is "Message body not downloaded" in the message window,
>and if I wait and wait and wait and wait some more,
>and if nothing further ever happens, then I begin to
>*suspect* that perhaps the news server responded that
>it didn't still have that message, although Opera
>never passes such a response back to me in any visible form.
>
>I should mention that in my news account Properties > Incoming
>I have no options checked, which means in particular that
>I do not wish to retain newsgroup message bodies at all,
>so every time I display a message, I presume that Opera
>will be going back to the news server for it
>and that Opera ought then to be able to know
>when a news server says "I can't retrieve that message"
>although I can never really get Opera to tell me that for sure.
Yep, Opera should know. The news server should get an error back,
probably 423 (that's what INN sends, it also uses "Bad article number"
in the optional comment field).
Of course if the server is really bad all bets are of... There have
been cases where servers just drops the connection when one attempts
to retreiving certain articles, but that's usually a sign of a
corrupted newsserver (disk crash or something has corrupted the
internal newsserver indexes).
sean
Use the 'read feeds' option - where all feeds are lumped together - delete/mark as read/etc. - and the number displayed to the right in the corresponding panel will reset to what's correct/current. Maybe - this is a bug.
--
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/
Matthew Winn
> Main issue was the claim that nntp was no meant
> for archive. It was and it worked wonderfully until Usenet
> became more mainstream and filled with the amount of nonsense
> that it is today.
NNTP was NOT meant for archiving. NNTP itself is purely a transport
mechanism and says nothing about what happens to news articles on
a server. However, the specification for NNTP (RFC977) refers to
RFC850, and that certainly does mention storage. In particular, it
says of the Expires header (my capitals):
This field is intended to be used to clean up articles
with a limited usefulness, OR TO KEEP IMPORTANT ARTICLES
AROUND FOR LONGER THAN USUAL. For example, a message
announcing an upcoming seminar could have an expiration
date the day after the seminar, since the message is not
useful after the seminar is over. SINCE LOCAL SITES HAVE
LOCAL POLICIES FOR EXPIRATION OF NEWS (DEPENDING ON
AVAILABLE DISK SPACE, FOR INSTANCE), users are discouraged
from providing expiration dates for articles unless there
is a natural expiration date associated with the topic.
This makes it quite clear that back in 1983 it was expected that news
articles would not be retained. There may have been some servers that
did retain articles for extended periods of time, but given the cost
of storage and the nature of discussions in the early USENET (which,
like today, were primarily electronic chattering, although without the
trolls or spammers) almost all servers had policies on expiring old
messages. That's why DejaNews had so much trouble trying to track
down archives of the early USENET: most servers had expired messages
within days or weeks of their arrival, so DejaNews had to try to find
old backup tapes and recover articles from those.
The early USENET was intended for bulletins and conversation, not as
a technical library and repository for reference documents. If you
have access to a news server that has long retention times, that's
great. But it's not specified that servers should retain messages for
any particular time, and you're certainly the first person I've ever
heard suggest that USENET was _intended_ to keep messages forever.
--
Matthew Winn
[If replying by email remove the "r" from "urk"]
djk
nntp allowed for articles to be archived. operational aspects of
any given server was the choice of the administration. as to the
content of posts, I guess it depends on what groups you read.
I came from a fidonet star to usenet mid-80's and there was
quite a bit of interesting stuff to read then, over the years
more people went to private servers, but still the same
protocol.
Matthew Winn
> your assumptions and conclusions are ridiculous in my opinion.
Your opinion doesn't change historical fact.
Read the specifications. Expiry of messages was standard behaviour.
It wasn't mandated, but given that storage was _very_ expensive back
then pretty much every administrator expired messages because there
no reason not to. Administrators who wanted to keep archives kept
them on tape, not on the server.
> nntp allowed for articles to be archived.
NNTP is a transport mechanism. It has nothing to do with archiving.
It makes no more sense to speak of NNTP allowing for articles to be
archived than it does to speak of SMTP allowing mail to be archived.
> operational aspects of
> any given server was the choice of the administration.
And they chose to expire articles. If, as you claim, administrators
kept online archives then why has it been so difficult to reconstruct
the early USENET? If you're right and USENET was intended for online
archiving then the reconstruction should have been easy.
> as to the
> content of posts, I guess it depends on what groups you read.
In the earliest days of USENET it was primarily a system for
distribution of bulletins. Why do you think it's called "news"?
> I came from a fidonet star to usenet mid-80's and there was
> quite a bit of interesting stuff to read then, over the years
> more people went to private servers, but still the same
> protocol.
Yes, there has been some interesting stuff, but as far as archiving
is concerned it was up to individuals to save what they thought they
might want to refer back to. It wasn't until years later that people
started to think "There's some good stuff on here; perhaps we should
start archiving it and making it available from a central resource".
This may come as a surprise to you, but just because you want to use
USENET in a particular way doesn't mean it was designed for that
purpose. USENET doesn't work in the way you'd like it to work and
it never has, and no amount of calling people ridiculous is going to
change that.
djk
assumptions and conclusions. my position that the
protocol was developed to allow archiving is true
and correct. just look at the header field design.
my assertion that archiving has taken place since
the beginning is true and correct.
your posts are grandiose bunk. entertaining in a
warped sense, like watching a dog chase it's own
tail.
John H Meyers
> Google is *not* a source of a true & complete archive.
No? It seems to still have everything I've ever posted,
going back to 1996 (several thousand items),
including all of what it inherited when it bought Deja[news],
and much which is even older than Dejanews
(where did they get it all?)
> There are still [NNTP?] servers out there
> that do have a fairly deep (aged) archive.
Yes, mine has thousands of current postings per group,
which is exactly why a fixed limit of only the most recent
251 posts per group seems not ideally flexible.
-[]-
Howard Brazee
On 6-Jul-2005, "John H Meyers" <jhme...@miu.edu> wrote:
> > Google is *not* a source of a true & complete archive.
>
> No? It seems to still have everything I've ever posted,
> going back to 1996 (several thousand items),
There's a switch some people or news readers put on messages to tell engines
such as Google to not pick up the messages. (I don't know what it is).
Jordi Kroon
wrote:
>
This lame header is X-No-Archive: Yes.
--
Jor
Using Opera's revolutionary e-mail client: < http://www.opera.com/mail/ >
djk
jordikATg...@127.0.0.1 says...
> This lame header is X-No-Archive: Yes.
>
>
their post from the google server after making it and sending it
out to the greater feed. this means posts exist on any reliable
good usenet server that you will not find on google. I can't
wait until google allows posters to *edit* prior posts without
sending the edited posts as new messages. people that use google
are fooling themselves, i think they are either too stupid to
set up a newsreader client or too lazy, or too cheap to have
access to a real news server, in any event, not too
intelligent. google and a lot of web based forums that for some
reason also feed right into usenet will make usenet a chatroom
like forum. illiterates will rule, if they don't already.
Bart Wakker
> your assumptions and conclusions are ridiculous in my opinion.
> nntp allowed for articles to be archived. operational aspects of
> any given server was the choice of the administration. as to the
Once more: NNTP is merely transport. It specifically contains
instructions about NOT to archive but not on archiving. The no-archive
flag only appeared in the mid 90s when dejanews was introduced (bought
later by google). Before that it was a non-issue because there was no
usenet software at all that would not expire ALL articles. The oldest
usenet software, A news, b news and c news, already had expiration and
in those days (mid 80's) any newsserver would expire pretty fast. I
ran a cnews server over UUCP at the time.
> content of posts, I guess it depends on what groups you read.
> I came from a fidonet star to usenet mid-80's and there was
> quite a bit of interesting stuff to read then, over the years
> more people went to private servers, but still the same
> protocol.
Long retention periods are something of the late 90s. I know only
private dedicated usenet servers that have no expiration. Any normal
server carrying the public groups has expiration. AFAIK google news,
is the only real archive around. I have never ever missed a message
except for those with a no-archive header, which I think should always
be respected.
Of course, the fixed limit of 251 headers is ridiculous. I'm using
emacs/GNUS so I don't really care however.
Bart Wakker
> well not only this but it seems any poster is allowed to delete
> their post from the google server after making it and sending it
> out to the greater feed. this means posts exist on any reliable
> good usenet server that you will not find on google. I can't
Another misconception on your side. Man, I've seldomly seen such
self-opinionated people, except on usenet :) :).
The possibility to CANCEL your own message has ALWAYS been present in
usenet. Yes there are servers that do not respect nor propagate the
cancel message, especially since there have been forged cancels, but
the original intention is that the poster can always "regret" his post
and then cancel his message. Usenet servers that would not respect
cancellation were condemned by most usenet administrators.
> wait until google allows posters to *edit* prior posts without
> sending the edited posts as new messages. people that use google
> are fooling themselves, i think they are either too stupid to
> set up a newsreader client or too lazy, or too cheap to have
> access to a real news server, in any event, not too
Not at all. I use my news client (a variety) maybe 30-60 minutes a
days, and have been since 1985. However, one has to use the right tool
for the right job. Until dejanews came in existance, there was no
archive. If you wondered about it you had to post the same question
that had been posted a thousand times before and hope that someone
would be so kind to answer (again). Since then it was possible to look
into the archive, thank god.
Usenet is NOT an efficient search protocol, however later additions
such as NOV have improved it somewhat. But expecting full retention on
a public newsserver to search into archives instead of google groups
is unrealistic. A newsreader is intended to follow and participate
into CURRENT discussions. Google-groups and maybe similar search
engines are intended and optimized for finding archived information.
Matthew Winn
> you continue to underlay your resonses with faulty
> assumptions and conclusions. my position that the
> protocol was developed to allow archiving is true
> and correct. just look at the header field design.
And what about the header field design makes you think NNTP was
developed with archiving in mind? Be specific. I've given ample
evidence that archiving was not the intention, ranging from the
fact that there were no archives in existence to the fact that
the standard specifically includes a header to allow messages to
temporarily override the administrators' expiry preferences.
Provide details, not insults.
> my assertion that archiving has taken place since
> the beginning is true and correct.
Apart from the fact that there were no online archives at the
beginning.
> your posts are grandiose bunk.
If they're bunk then why is it that you can't refute a single point?
P.S. Learn to quote.
djk
cancel is dependent on the administration of the server. all the
other stuff is just a rehash. i'll end by saying to you and the
other guy and anybody else, *believe what you want to believe*
OK? enough said in this thread, let it be. [move along, nothing
to see here.]
>
djk
o*@matthewwinn.me.urk says...
> P.S. Learn to quote.
>
>
what? meaningless drivel? I don't think it is worth the
groups time. Again, you present just a rehash, like your buddy.
I don't buy for a minute what you present. So let's agree
to disagree. I'd end by saying "learn to read", but it would
be useless. Good luck.
djk
allows for the archive of posts. end of post. end of discussion.
please move on. good luck.
>
Matthew Winn
On Thu, 07 Jul 2005 21:57:26 GMT, djk <d...@spam.operamail.com> wrote:
> In article <slrndcpmu5.odd.o*@mwinn.powernet.co.uk>,
> o*@matthewwinn.me.urk says...
> > P.S. Learn to quote.
> >
> >
> what? meaningless drivel? I don't think it is worth the
> groups time. Again, you present just a rehash, like your buddy.
When are you going to tell us which part of the header information
shows that archiving was designed in from the start? I hope you're
not referring to the X-No-Archive header (though I suspect you are)
because that was added by DejaNews when they set up the first public
USENET archive a decade or so ago, and the reason they added it was
because of some people's privacy concerns when they found that, for
the first time, all articles posted to USENET were being archived.
It's not a standard header, as the leading "X-" makes clear.
> I don't buy for a minute what you present.
Why not? Point out the errors in my (and others') arguments.
You haven't presented any evidence at all in defence of your position.
All you've done is insult people and make assertions that you're
unable to support. Most of us here are techies, and you're going to
have to do better than that if you want us to stop laughing at you.
It's been fun, but I'd still like to know where you got this idea that
USENET was designed for archiving from the very beginning.
djk
here is a clue for you and your buddy,
the mere fact that the protocol header design allows for the
expiration of articles posted means that the protocol was
designed with the ability to archive. the admin of the server
can establish how long the article lasts on the server, thus has
the ability (given by the nttp protocol header design) to
archive.
now, if either of you don't understand that,
continue on with the mindless drivel. I'm done
trying to get it through to you knuckle heads.
Rijk van Geijtenbeek
> read this,
> here is a clue for you and your buddy,
Please try to keep conversations in these newsgroups civil, even if you
can't come to an understanding.
--
The Web is a procrastination apparatus: | Rijk van Geijtenbeek
It can absorb as much time as | Documentation & QA
is required to ensure that you | Opera Software ASA
won't get any real work done. - J.Nielsen
|http://my.opera.com/Rijk/journal
Richard Grevers
> read this,
> here is a clue for you and your buddy,
>
> the mere fact that the protocol header design allows for the
> expiration of articles posted means that the protocol was
> designed with the ability to archive.
Actually, the fact that the No-Archive header is X-No-Archive demonstrates
that it is NOT part of the protocol (other than the protocol allowing for
aftermarket headers commencing with X).
But heading back to where we started, the way news articles are propagated
means that even if you have a news server with a deep archive of a group,
there is no guarantee that its record will be complete. In fact I believe
that Deja/google monitor multiple feeds in order to reduce this problem.
I believe that there is a strong analogy between usenet and newspapers.
The vast majority of people only want today's paper. Occasionally they
want to keep ana rticle so they clip it out and store it somewhere safe,
then trow away the rest. If you want to find an old item, any reasonable
person would go to the library and search for it via their index of back
issues (and the library lets you make copies) rather than go to the
newspaper office and buy six months worth of old newspapers to search
through.
Hence I don't see wanting to download thousands of messages from a
newsgroup in order to conduct a search as rational behaviour (nor for that
matter is wanting to read discussions about Opera 7.1 rather than 8.01).
Of course one might accumulate a large repository of messages over time -
I see that I have 362 days worth of opera.general, which is probably
overdue for a cleanout. In fact, I would rate purging control (and locking
of messages you want to keep) a far higher priority than an option for how
many messages to download. I do believe the latter should be done, but
more because 250 is often too high a number, e.g. if I know I am up to
date with a group via another computer and I only want today's new posts.
Richard Grevers
djk
Matthew Winn
On Fri, 08 Jul 2005 16:41:59 GMT, djk <d...@spam.operamail.com> wrote:
> read this,
> here is a clue for you and your buddy,
>
> the mere fact that the protocol header design allows for the
> expiration of articles posted means that the protocol was
> designed with the ability to archive.
No it doesn't. It was designed with the ability _not_ to archive.
There are only two headers you could be talking about:
(1) The expiry date. This allows for articles to override the
default expiry mechanism, so they can be expired more rapidly than
usual _or_ so they can be retained longer than usual. It was a way
of deferring expiry for a while. That's a completely different thing
from archiving.
(2) The X-No-Archive header. That's a recent addition, introduced
when the first USENET archives were created last decade.
So where's this mystical archiving header of which you speak?
> the admin of the server
> can establish how long the article lasts on the server, thus has
> the ability (given by the nttp protocol header design) to
> archive.
And where did they get the disk space? We're talking about storing
several megabytes of data every day on the storage devices of twenty
years ago. Maintaining the sort of archive you're referring to would
have cost thousands of dollars a week, hundreds of thousands a year.
You're talking about a time when a 40MB drive was large both in terms
of storage space and physical size. Back then you expired EVERYTHING
the moment it wasn't needed. Archives were on tape. Most people
wouldn't have entertained even the concept of an online archive back
then. Waste precious space on old data? Unthinkable!
Find an example of a site that did what you say. If, as you claim,
archiving was a standard feature then the network should be packed
full of them, so let's see some evidence of their existence. Where's
the proof? If all you can do is hurl insults then you make it clear
to all our readers that you're wrong and you know you're wrong.
> now, if either of you don't understand that,
> continue on with the mindless drivel. I'm done
> trying to get it through to you knuckle heads.
But we're having fun watching you try to defend your position when
you can't come up with even a shred of fact to support it. Proof by
insult has never been considered a credible debating tactic, though
it's certainly entertaining to watch.
djk
specifically. I was addressing the condition of being able
to routinely expire articles at the admins choosing, thusly
deciding on the size of the archive the news server offers.
An archive, by definition, can be any length of past time.
By design the nntp header allows for the archive of posts.
To everybody, I'm done. Carry on and good luck.
>