Re: All languages are equally fit
Joachim Pense
> On 2009-11-02, John Atkinson wrote:
>
>> Joachim Pense wrote:
>>> Peter T. Daniels (in sci.lang):
>
>>>> I don't suppose there's a verb "stoppen," outside the routines of Mel
>>>> Brooks and Carl Reiner?
>>>>
>>> "Stoppen" is a common colloquial word in German and has been so for
>>> decades. "Wie kann man die Wirtschaftskrise stoppen?"
>>>
>> Yes, I'm surprised Peter hasn't come across it in his reading, since I
>> have, and I have much less German than he does.
>
> It's not used in academic writing, is it?
>
>
No.
Joachim
Peter T. Daniels
I was wondering how Adam had managed to add a 1001st posting to that
thread. Here, though, what would have been the 1002nd started a new
thread.
Ruud Harmsen
<gram...@verizon.net>: in sci.lang:
>I was wondering how Adam had managed to add a 1001st posting to that
>thread. Here, though, what would have been the 1002nd started a new
>thread.
He posted with a stand-alone Usenet program running on Ubuntu Linux.
The silly 1000 postings limitation is imposed by Google, not by Usenet
protocols.
I can't repeat it often enough: Google is NOT Usenet, just as the WWW
is NOT the internet.
--
Ruud Harmsen, http://rudhar.com
António Marques
Google Groups is not 'Usenet' in many more ways than the WWW is not the
Internet.
Maybe 'Google Groups is not Usenet, just as Google Web Search is not the WWW'?
Peter T. Daniels
> Mon, 14 Dec 2009 13:20:43 -0800 (PST): "Peter T. Daniels"
>
> >I was wondering how Adam had managed to add a 1001st posting to that
> >thread. Here, though, what would have been the 1002nd started a new
> >thread.
>
> He posted with a stand-alone Usenet program running on Ubuntu Linux.
> The silly 1000 postings limitation is imposed by Google, not by Usenet
> protocols.
>
> I can't repeat it often enough: Google is NOT Usenet, just as the WWW
> is NOT the internet.
So does everyone who does not use google groups see these (now) four
postings as the 1002nd through 1005th items in the main thread?
António Marques
Everyone not using google groups is probably using something that doesn't
ascribe numbers to messages. Don't remeber when you used Netscape?
Ruud Harmsen
<gram...@verizon.net>: in sci.lang:
>> I can't repeat it often enough: Google is NOT Usenet, just as the WWW
>> is NOT the internet.
>
>So does everyone who does not use google groups see these (now) four
>postings as the 1002nd through 1005th items in the main thread?
They aren't counted in Usenet per se, all messages are separate
entities but have references to earlier messages.
Peter T. Daniels
Years ago, and dial-up; the messages were sent to me, I didn't go and
get them.
It's not a question of numbering messages, it's a question of counting
them.
Joachim Pense
But these are not complete, so you cannot use the references to find out the
number, or even the depth.
Joachim
Joachim Pense
> Mon, 14 Dec 2009 13:20:43 -0800 (PST): "Peter T. Daniels"
> <gram...@verizon.net>: in sci.lang:
>
>>I was wondering how Adam had managed to add a 1001st posting to that
>>thread. Here, though, what would have been the 1002nd started a new
>>thread.
>
> He posted with a stand-alone Usenet program running on Ubuntu Linux.
> The silly 1000 postings limitation is imposed by Google, not by Usenet
> protocols.
>
Do they also have a time limitation? I think they must have something like
that to avoid people replying to threads from 1991.
Joachim
LEE Sau Dan
>>> So does everyone who does not use google groups see these (now)
>>> four postings as the 1002nd through 1005th items in the main
>>> thread?
>>
>> They aren't counted in Usenet per se, all messages are separate
>> entities but have references to earlier messages.
Joachim> But these are not complete, so you cannot use the
Joachim> references to find out the number, or even the depth.
There is no unique "message number". Each NNTP server assigns different
message numbers to articles. Usually, it assigns the numbers
consecutively according to the time the message is received by the
server. However, since messages can arrive in different orders at
different servers, due to the decentralized, distributed nature of the
NNTP network, there is no single and unique way to assign numbers to the
whole pool of messages. Can't they simply assign the numbers according
to the "Date:" header? No. That header gives the timestamp for the
creation of a message, which is not the same as the time that it arrives
at an NNTP server. Messages aren't received in ascending order of their
"Date:" field, because *relayed* messages can arrive at an NNTP server
later than more recently created messages, and you won't know ahead of
time how many delayed and relayed messages you're going to get.
--
Lee Sau Dan 李守敦 ~{@nJX6X~}
E-mail: dan...@informatik.uni-freiburg.de
Home page: http://www.informatik.uni-freiburg.de/~danlee
Adam Funk
> On Dec 15, 2:40 am, Ruud Harmsen <r...@rudhar.eu> wrote:
>> Mon, 14 Dec 2009 13:20:43 -0800 (PST): "Peter T. Daniels"
>> <gramma...@verizon.net>: in sci.lang:
>>
>> >I was wondering how Adam had managed to add a 1001st posting to that
>> >thread. Here, though, what would have been the 1002nd started a new
>> >thread.
>>
>> He posted with a stand-alone Usenet program running on Ubuntu Linux.
>> The silly 1000 postings limitation is imposed by Google, not by Usenet
>> protocols.
>>
>> I can't repeat it often enough: Google is NOT Usenet, just as the WWW
>> is NOT the internet.
You can lead a horse to water...
> So does everyone who does not use google groups see these (now) four
> postings as the 1002nd through 1005th items in the main thread?
How many messages you see in the thread depends on the news client,
the server retention, etc., but Google's thread-breaks are just its
own misleading, clunky invention.
Also, the USENET really consists of a bunch of individual messages.
The *news client* assembles threads based on the References (and in
many cases Subject) headers for the user's convenience.
--
Nobody ever went broke underestimating the taste of
the American public. [Mencken]
Adam Funk
Your news client fetched them (and it should've been more noticeable
on dial-up). The messages only get "pushed" as far as your news
provider's server. When you click on a group in Netscape (for
example), it queries the server for the number of new messages since
the highest message number [see below] you got last time; then it
fetches the basic headers for all the new messages and displays these.
(It also updates the "highest number" for that group.) When you click
on an individual message in the list for a group, it fetches the
remaining headers and the message body and displays the message.
> It's not a question of numbering messages, it's a question of counting
> them.
Message do have sequential numbers within a group on a particular news
server, but there is no correspondence between the same message's
number on different servers, except that if message A is much newer
than message B, then A *probably* has a higher number on most servers
than B. But these numbers are assigned by a server in the sequence in
which it receives them from clients and other servers, which does not
follow timestamp order.
If two servers started carrying the same group at different times, the
same message could easily be 101 on one server and 234567 on another.
This is why if you change news providers, the read and unread status
of the messages on your list already gets lost.
--
"Mrs CJ and I avoid clichés like the plague."
Artur Jachacy
I checked in Mozilla Thunderbird, which I don't normally use for Usenet,
and it says the total for this thread is 1334 (now probably 1335).
A.
--
Gridneff: So the point is to get the message across,
without saying in so many words: You stupid fucking morons,
you're learning fucking elf languages!
Pan for Windows (beta) - <http://panbuilds.googlepages.com>
LEE Sau Dan
Artur> I checked in Mozilla Thunderbird, which I don't normally use
Artur> for Usenet, and it says the total for this thread is 1334
Artur> (now probably 1335).
What a meaningless number.
Do you know the retention policy of the NNTP server you're using? Maybe
this thread has got 10000 messages, but your NNTP server is only
retaining the most recent 1335 of them. Have you ever investigated into
this issue and collected evidences to refute what I just said?
Another question: do you know how Thunderbird group the messages into a
thread (actually, it's a tree)? Some NNTP clients follow strictly on
the "References:" header only, while others would be have more
aggressive heuristics and group also messages with identical or similar
(e.g. up to prefix "Re:") subject lines into the same thread. The
threading algorithm varies from client to client. So, have you studied
Thunderbird document or source code to know which algorithm it uses?
Peter T. Daniels
wrote:
> >>>>> "Artur" == Artur Jachacy <arturj.use...@gmail.com> writes:
>
> Artur> I checked in Mozilla Thunderbird, which I don't normally use
> Artur> for Usenet, and it says the total for this thread is 1334
> Artur> (now probably 1335).
>
> What a meaningless number.
>
> Do you know the retention policy of the NNTP server you're using? Maybe
> this thread has got 10000 messages, but your NNTP server is only
> retaining the most recent 1335 of them. Have you ever investigated into
> this issue and collected evidences to refute what I just said?
>
> Another question: do you know how Thunderbird group the messages into a
> thread (actually, it's a tree)? Some NNTP clients follow strictly on
> the "References:" header only, while others would be have more
> aggressive heuristics and group also messages with identical or similar
> (e.g. up to prefix "Re:") subject lines into the same thread. The
> threading algorithm varies from client to client. So, have you studied
> Thunderbird document or source code to know which algorithm it uses?
Well, then I guess google groups does pretty well, because every
message appears "treed" directly under the message it replies to, and
there's no reason to suppose that old messages get chopped off the
tops of threads.
LEE Sau Dan
Peter> Well, then I guess google groups does pretty well, because
Peter> every message appears "treed" directly under the message it
Peter> replies to, and there's no reason to suppose that old
Peter> messages get chopped off the tops of threads.
Are you sure they don't do so?
Peter T. Daniels
wrote:
> >>>>> "Peter" == Peter T Daniels <gramma...@verizon.net> writes:
>
> Peter> Well, then I guess google groups does pretty well, because
> Peter> every message appears "treed" directly under the message it
> Peter> replies to, and there's no reason to suppose that old
> Peter> messages get chopped off the tops of threads.
>
> Are you sure they don't do so?
I have never yet encountered a thread (and we've had now three 1000-
message threads recently) where the oldest one, the original one, fell
off the back end.
Of course a poster can mark their posting to not be archived, and then
after a week it goes away.
Artur Jachacy
>>>>>> "Artur" == Artur Jachacy <arturj...@gmail.com> writes:
>
> Artur> I checked in Mozilla Thunderbird, which I don't normally use
> Artur> for Usenet, and it says the total for this thread is 1334
> Artur> (now probably 1335).
>
> What a meaningless number.
>
> Do you know the retention policy of the NNTP server you're using? Maybe
> this thread has got 10000 messages, but your NNTP server is only
> retaining the most recent 1335 of them. Have you ever investigated into
> this issue and collected evidences to refute what I just said?
It's my own server so I set the retention policy myself.
> Another question: do you know how Thunderbird group the messages into a
> thread (actually, it's a tree)? Some NNTP clients follow strictly on
> the "References:" header only, while others would be have more
> aggressive heuristics and group also messages with identical or similar
> (e.g. up to prefix "Re:") subject lines into the same thread. The
> threading algorithm varies from client to client. So, have you studied
> Thunderbird document or source code to know which algorithm it uses?
No, I haven't and I don't intend to. However, I can see that it doesn't
do what you said because the count doesn't include some messages which
were probably in answer to the original post but don't include the
appropriate References: header.
PaulJK
It indeed depends on what kind of agent one uses. Some Usenet
agents have selectable options that allows several different ways
of viewing the threads.
When you display the threaded tree structure of the posts then the
post submitted by Google as "new" thread would indeed be shown
at the root of a brand new thread because that's how Google
submits it. That is irrespective whether the subject text is changed
or not. Equally, you can have threaded posts with different subject
texts shown as part of the same tree. The information about where
the post belongs within the thread tree is contained in each post's
header.
For the last several years I have been using a different option.
I view all posts with new and unread arrivals highlighted, currently
on my hard drive, sorted by NZDT submit time within the alphabetically
sorted subject texts. I am not interested in any ISP's serial numbers
and I also ignore the thread info in the posts' headers.
All posts with the same subject text are sorted chronologically by
their time of submission so I am completely unaware of this particular
problem caused by Google. On the other hand, when the responder
changes the subject text, I don't immediately see the relationship
between the 'question' and 'respond' posts. That is, not until I actually
open it and read it.
Regarding the serial message numbers: They are local
arbitrary sequence numbers assigned by each Usenet server as
it receives them over the internet highway. There are thousands
of servers around the world and they maintain mutually
incompatible sequence numbers. Their use is purely
administrational between each server and the agents reading
from its Usenet database. Every time an agent connects to
the (non-Google) server to read a particular group, it sends
it the last received serial number for that group. The server
responds by sending it all messages with higher numbers,
i.e. the ones that have arrived since the agent connected
last time. No messages already received need to be
retransmitted to the user's agent. That is how it has been
done for decades, not until Google introduced its idiotic
mindless system that often keeps resending you and
redisplaying lot of stuff you've already seen before.
pjk
PaulJK
There is no central computer to assing world wide serial
numbers. Each of the many thousands of Usenet servers
might receive messages in different order than other servers.
If there are any serial numbers assigned, they are local
affairs between each server and agents connected to it.
These are usually assigned in ascending order per each
user group and serve the purpose of finding quickly what
posts each agent has not yet received.
The serial numbers assigned to each thread is a Google
invention with its own unique plethora of bugs.
pjk
PaulJK
No, you did go out requesting the messages. It only may have
seem to you like you didn't. The request was probably sent
as soon as started the Usenet agent.
The agents would *always* request the messages first with
a particular command containing a group name and the next
msg serial number to be sent.
> It's not a question of numbering messages, it's a question of counting
> them.
?
pjk
PaulJK
There is absolutely no world wide best by date.
Each Usenet server has its own policy on expiring old messages.
However, absolutely nothing stops you from responding to
a thread from 1991 if post of such thread still sit on your computer.
The message header of your response will quote message(s)
it is supposed to follow. If such message is received by another
agent who still has such messages on his computer then it
shows your response correctly slotted in the tree structure.
pjk
António Marques
> Every time an agent connects to
> the (non-Google) server to read a particular group, it sends
> it the last received serial number for that group. The server
> responds by sending it all messages with higher numbers,
> i.e. the ones that have arrived since the agent connected
> last time. No messages already received need to be
> retransmitted to the user's agent. That is how it has been
> done for decades, not until Google introduced its idiotic
> mindless system that often keeps resending you and
> redisplaying lot of stuff you've already seen before.
What do you mean? Does google offer some king of NNTP [look-alike]?
I don't see that displaying numbers next to messages in the web thread
view is particularly damning. Iinm, a message will have a different
number according to the type of list (threaded vs by date). It's light
years away from offering a reasonable user experience, but I don't see
the competition coming up with something better. In fact, a usable
thread reader is something that is long overdue, whether on the web or
in standalone clients. Decades of this and nobody's come up with
something nice to use.
PaulJK
>>>>>> "Peter" == Peter T Daniels <gram...@verizon.net> writes:
>
> Peter> Well, then I guess google groups does pretty well, because
> Peter> every message appears "treed" directly under the message it
> Peter> replies to, and there's no reason to suppose that old
> Peter> messages get chopped off the tops of threads.
>
> Are you sure they don't do so?
Apart from that, Peter would have find it a bit disturbing
if he 'saw' messages that Google displays hours or days
late as well as messages that it manages to loose completely.
But since he doesn't use a good agent in parallel he doesn't
get to see them.
pjk
PaulJK
> On 16 Dez, 04:29, "PaulJK" <paul.kr...@paradise.net.nz> wrote:
>> Every time an agent connects to
>> the (non-Google) server to read a particular group, it sends
>> it the last received serial number for that group. The server
>> responds by sending it all messages with higher numbers,
>> i.e. the ones that have arrived since the agent connected
>> last time. No messages already received need to be
>> retransmitted to the user's agent. That is how it has been
>> done for decades, not until Google introduced its idiotic
>> mindless system that often keeps resending you and
>> redisplaying lot of stuff you've already seen before.
>
> What do you mean? Does google offer some king of NNTP [look-alike]?
What do you mean what do I mean?
Of course Google functions as a Usenet server. AND simultaneously
it also functions as an agent with fixed properties. You are given
access to the 'agent' functions via its web page interface.
You have very little control over its functionality.
> I don't see that displaying numbers next to messages in the web thread
> view is particularly damning. Iinm, a message will have a different
> number according to the type of list (threaded vs by date). It's light
> years away from offering a reasonable user experience, but I don't see
> the competition coming up with something better. In fact, a usable
> thread reader is something that is long overdue, whether on the web or
> in standalone clients. Decades of this and nobody's come up with
> something nice to use.
Pinch me, I must be dreaming.... :-)
pjk
Joachim Pense
> Joachim Pense wrote:
>> Ruud Harmsen (in sci.lang):
>>
>>> Mon, 14 Dec 2009 13:20:43 -0800 (PST): "Peter T. Daniels"
>>> <gram...@verizon.net>: in sci.lang:
>>>
>>>> I was wondering how Adam had managed to add a 1001st posting to that
>>>> thread. Here, though, what would have been the 1002nd started a new
>>>> thread.
>>>
>>> He posted with a stand-alone Usenet program running on Ubuntu Linux.
>>> The silly 1000 postings limitation is imposed by Google, not by Usenet
>>> protocols.
>>>
>>
>> Do they also have a time limitation? I think they must have something
>> like that to avoid people replying to threads from 1991.
>> Joachim
>
> There is absolutely no world wide best by date.
>
> Each Usenet server has its own policy on expiring old messages.
> However, absolutely nothing stops you from responding to
> a thread from 1991 if post of such thread still sit on your computer.
>
Sure I could do such a thing if I wanted to. Only, Google Groups, displaying
all the historical posts, looks like an invitation to respond to very old
threads, including ones that became history. I did not check if that is
actually done, but I think if I were Google, I'd put in a mechanism into
the user interface to stop people from doing that using Google Groups.
Joachim
António Marques
> António Marques wrote:
>> On 16 Dez, 04:29, "PaulJK"<paul.kr...@paradise.net.nz> wrote:
>>> Every time an agent connects to
>>> the (non-Google) server to read a particular group, it sends
>>> it the last received serial number for that group. The server
>>> responds by sending it all messages with higher numbers,
>>> i.e. the ones that have arrived since the agent connected
>>> last time. No messages already received need to be
>>> retransmitted to the user's agent. That is how it has been
>>> done for decades, not until Google introduced its idiotic
>>> mindless system that often keeps resending you and
>>> redisplaying lot of stuff you've already seen before.
>>
>> What do you mean? Does google offer some king of NNTP [look-alike]?
>
> What do you mean what do I mean?
>
> Of course Google functions as a Usenet server. AND simultaneously
> it also functions as an agent with fixed properties. You are given
> access to the 'agent' functions via its web page interface.
> You have very little control over its functionality.
How does google function as a usenet server?
>> I don't see that displaying numbers next to messages in the web thread
>> view is particularly damning. Iinm, a message will have a different
>> number according to the type of list (threaded vs by date). It's light
>> years away from offering a reasonable user experience, but I don't see
>> the competition coming up with something better. In fact, a usable
>> thread reader is something that is long overdue, whether on the web or
>> in standalone clients. Decades of this and nobody's come up with
>> something nice to use.
>
> Pinch me, I must be dreaming.... :-)
MSOE, usable? For Chrisssake.
António Marques
> LEE Sau Dan wrote:
>>>>>>> "Peter" == Peter T Daniels<gram...@verizon.net> writes:
>>
>> Peter> Well, then I guess google groups does pretty well, because
>> Peter> every message appears "treed" directly under the message it
>> Peter> replies to, and there's no reason to suppose that old
>> Peter> messages get chopped off the tops of threads.
>>
>> Are you sure they don't do so?
>
> Apart from that, Peter would have find it a bit disturbing
> if he 'saw' messages that Google displays hours or days
> late as well as messages that it manages to loose completely.
Again, I don't see that google is especially offending in this regard. But
then again, I've never met a Usenet provider that worked fine (current one
almost cuts it).
> But since he doesn't use a good agent in parallel he doesn't
> get to see them.
Again, what *web* readers are better than google?
António Marques
And they did, as has been mentioned on occasion. You fellow languageman from
the Switz regularly starts new threads when he becomes unable to keep the
old ones going.
Marco Pagliero
> Sure I could do such a thing if I wanted to. Only, Google Groups, displaying
> all the historical posts, looks like an invitation to respond to very old
> threads, including ones that became history. I did not check if that is
> actually done, but I think if I were Google, I'd put in a mechanism into
> the user interface to stop people from doing that using Google Groups.
month old.
Marco P
Peter T. Daniels
I think I would know if I saw replies to a message that I didn't see!
Maybe once in my 5 1/2 years using google groups there's been a system-
wide delay that held messages up for as much as 12 hours. I'd say that
just about without exception, they appear quickly enough that it's
like having an IM conversation.
Peter T. Daniels
Cardinal vs. ordinal -- what does "ascribe numbers to" mean?
Peter T. Daniels
> Ruud Harmsen wrote:
> > Tue, 15 Dec 2009 05:10:46 -0800 (PST): "Peter T. Daniels"
I can view a thread in two ways, either "sort by date" (which uses the
timestamp in the message), and here the serial numbers, from 1 to
1000, don't change (unless a message got delayed somewhere, I
suppose); or "sort by reply" (which puts each message directly under
the one it replies to; the serial numbers change as replies to earlier
branches appear). I use the latter way except when there are hundreds
of messages and replies might be anywhere in the thread. (I can see 26
author-lines of messages at once, the rest of the screen being taken
up by IE headers and such). Either way, messages I haven't looked at
are in bold. When I click on an author-name, a group of ten messages
(from xx1 to xx0) comes into view, and I can't start typing a reply
and then go to look at a message not within that decade (I have to
send the reply, or it will be deleted unsent).
Adam Funk
I forgot to mention that messages do have unique [see below] IDs
generated either by the user's news client or by the first NNTP server
that handles the message (it goes in the Message-ID header). In the
"forwarding" process (NNTP IHAVE command), other servers use these to
determine whether they already have each message.
A well-formed MID has basically the same syntax as an e-mail address,
<local...@host.example.com>, where the local part is supposed to be
uniquely generated (usually from some combination of hashes,
timestamps, and PIDs, rather than a sequential counter). The IDs are
not sequential or similarly meaningful. If a message has the same ID
as one already on the server, the server will refuse to take it, and
you will get an error message.
--
A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Maybe because some people are too annoyed by top-posting.
Q: Why do I not get an answer to my question(s)?
António Marques
Specifically, the MID cannot be used as a number becuse there is just no way
to turn it into a (usable) unique number.
PaulJK
> PaulJK wrote, on 16-12-2009 05:19:
>> António Marques wrote:
>>> On 16 Dez, 04:29, "PaulJK"<paul.kr...@paradise.net.nz> wrote:
>>>> Every time an agent connects to
>>>> the (non-Google) server to read a particular group, it sends
>>>> it the last received serial number for that group. The server
>>>> responds by sending it all messages with higher numbers,
>>>> i.e. the ones that have arrived since the agent connected
>>>> last time. No messages already received need to be
>>>> retransmitted to the user's agent. That is how it has been
>>>> done for decades, not until Google introduced its idiotic
>>>> mindless system that often keeps resending you and
>>>> redisplaying lot of stuff you've already seen before.
>>>
>>> What do you mean? Does google offer some king of NNTP [look-alike]?
>>
>> What do you mean what do I mean?
>>
>> Of course Google functions as a Usenet server. AND simultaneously
>> it also functions as an agent with fixed properties. You are given
>> access to the 'agent' functions via its web page interface.
>> You have very little control over its functionality.
>
> How does google function as a usenet server?
By collecting/assembling usenet articles within their own
database and dishing/serving them out upon request.
Isn't that obvious?
What do you think the word 'server' means in this context.
That is more or less what zillions of other usenet servers do.
What Google does differently is that they put their own idiosyncratic
agent in front of it and allow users access to their database
through this agent via internet http protocol, specifically html pages.
That allows unsophisticated users without an agent access to
usenet via http protocol of their browser. But it also forces the
limited functionality of their agent on all their google group users.
They also never retire old and very old items from their database
which, as Joachim mentions, often results in users responding
to ancient articles.
pjk
PaulJK
> On Dec 15, 11:52 pm, "PaulJK" <paul.kr...@paradise.net.nz> wrote:
>> Peter T. Daniels wrote:
I guess Antonio meant to say 'assign numbers'.
In general, Usenet doesn't assign message numbers or count
messages, there is no central point to assign unique numbers or
do the counting. What Google does, be it counting or number
assignement, between their own message database and the
end user is not part of Usenet system.
pjk
PaulJK
Remember that these serial numbers are assigned by Google
when they receive the messages in their database, they are
not seen by non-Google group users and they have no meaning
outside the Google group environment.
> don't change (unless a message got delayed somewhere, I suppose);
> or "sort by reply" (which puts each message directly under
> the one it replies to;
A reply cannot be always *directly* under the post it replies to.
Quite often you have more than one replies to a particular post.
You need two dimensions to represent such structure.
There are many agents available that can display message IDs
connected by lines not unlike the Window Explorer displays
the hierarchical structure of directories with files and directories
within them.
> the serial numbers change as replies to earlier
> branches appear). I use the latter way except when there are hundreds
> of messages and replies might be anywhere in the thread. (I can see 26
> author-lines of messages at once, the rest of the screen being taken
> up by IE headers and such). Either way, messages I haven't looked at
> are in bold. When I click on an author-name, a group of ten messages
> (from xx1 to xx0) comes into view,
that's what leads to unncessary retransmisson of a lot of
already received data and slows you down.
I prefer using agents that receive all new messages only once
(by that I mean messages I've never ever received before)
and does it in the background as soon as I log on.
By the time I start browsing hierarchies of the posts, be it
sequentially or randomly, the agent reads all messages off my
hard disk which is lightningly fast. I can ask to redisplay the
hierarchy in many different sorting orders and it gets done and
displayed in milliseconds. I am well familiar with Google groups
and I know that doing that in Google is just one big pain in the
arse.
> and I can't start typing a reply
> and then go to look at a message not within that decade (I have to
> send the reply, or it will be deleted unsent).
That is all a consequence the Google 'agent' not being
resident on your machine. No posts are filed away on your
machine and you have only a very narrow window (10 wide
in your case) through which you view messages in the
Google's database. Whenever you ask for the next lot or
a different sorting order the whole bundle is resend again
regardless whether you had it on your machine already or
not.
pjk
PaulJK
Remember that these serial numbers are assigned by Google
when they receive the messages in their database, they are
not seen by non-Google group users and they have no meaning
outside the Google group environment.
> don't change (unless a message got delayed somewhere, I suppose);
> or "sort by reply" (which puts each message directly under
> the one it replies to;
A reply cannot be always *directly* under the post it replies to.
Quite often you have more than one replies to a particular post.
You need two dimensions to represent such structure.
There are many agents available that can display message IDs
connected by lines not unlike the Window Explorer displays
the hierarchical structure of directories with files and directories
within them.
> the serial numbers change as replies to earlier
> branches appear). I use the latter way except when there are hundreds
> of messages and replies might be anywhere in the thread. (I can see 26
> author-lines of messages at once, the rest of the screen being taken
> up by IE headers and such). Either way, messages I haven't looked at
> are in bold. When I click on an author-name, a group of ten messages
> (from xx1 to xx0) comes into view,
that's what leads to unncessary retransmisson of a lot of
already received data and slows you down.
I prefer using agents that receive all new messages only once
(by that I mean messages I've never ever received before)
and does it in the background as soon as I log on.
By the time I start browsing hierarchies of the posts, be it
sequentially or randomly, the agent reads all messages off my
hard disk which is lightningly fast. I can ask to redisplay the
hierarchy in many different sorting orders and it gets done and
displayed in milliseconds. I am well familiar with Google groups
and I know that doing that in Google is just one big pain in the
arse.
> and I can't start typing a reply
> and then go to look at a message not within that decade (I have to
> send the reply, or it will be deleted unsent).
That is all a consequence the Google 'agent' not being
Peter T. Daniels
Obviously, the replies to the first reply go directly under that one,
etc., so we have a nice series of indents (two dimensions). The
structure is alweays clear.
> There are many agents available that can display message IDs
> connected by lines not unlike the Window Explorer displays
> the hierarchical structure of directories with files and directories
> within them.
>
> > the serial numbers change as replies to earlier
> > branches appear). I use the latter way except when there are hundreds
> > of messages and replies might be anywhere in the thread. (I can see 26
> > author-lines of messages at once, the rest of the screen being taken
> > up by IE headers and such). Either way, messages I haven't looked at
> > are in bold. When I click on an author-name, a group of ten messages
> > (from xx1 to xx0) comes into view,
>
> that's what leads to unncessary retransmisson of a lot of
> already received data and slows you down.
>
> I prefer using agents that receive all new messages only once
> (by that I mean messages I've never ever received before)
> and does it in the background as soon as I log on.
> By the time I start browsing hierarchies of the posts, be it
> sequentially or randomly, the agent reads all messages off my
> hard disk which is lightningly fast. I can ask to redisplay the
There are no messages on my hard disk (except on the rare occasions
when I Save one).
> hierarchy in many different sorting orders and it gets done and
> displayed in milliseconds. I am well familiar with Google groups
> and I know that doing that in Google is just one big pain in the
> arse.
As I said, there are two displays, and switching between them takes
only a moment.
Joachim Pense
> António Marques wrote:
>> How does google function as a usenet server?
>
> By collecting/assembling usenet articles within their own
> database and dishing/serving them out upon request.
>
> Isn't that obvious?
> What do you think the word 'server' means in this context.
>
I would think Google run a real usenet NNTP server to interface to the
usenet, or maybe a number of them, and feed their database from it, and
also feed the usenet through it. But I don't know any details.
> That is more or less what zillions of other usenet servers do.
"More or less" for sure, but that's the trivial part.
> They also never retire old and very old items from their database
> which, as Joachim mentions, often results in users responding
> to ancient articles.
Someone mentioned in this thread that Google doesn't allow you to answer to
articles older than a month.
Joachim
Joachim Pense
> When I click on an author-name, a group of ten messages
> (from xx1 to xx0) comes into view, and I can't start typing a reply
> and then go to look at a message not within that decade (I have to
> send the reply, or it will be deleted unsent).
Of course you could work with a second browser window or tab to view other
messages while composing a new one.
Joachim
António Marques
Obvious it is, but that's not what they do. They do not serve articles upon
request, they only serve web pages which happen to be usable as a GUI to
access the content of articles.
What's next, calling a webmail interface a mail server? A mail server is
something which implements POP and/or IMAP and/or SMTP.
> What do you think the word 'server' means in this context.
An NNTP server, which is what you were comparing google to, can only mean
one thing: a server implementing the NNTP protocol. I read your paragraph as
implying that google did provide some quasi-NNTP interface.
> That is more or less what zillions of other usenet servers do.
>
> What Google does differently is that they put their own idiosyncratic
> agent in front of it and allow users access to their database
> through this agent via internet http protocol, specifically html pages.
Not quite; what they do differently is that their NNTP server, if it exists,
is only accessible internally to their web (server) application. That that
they provide a web UI to do 'usenet stuff' is a separate thing.
> That allows unsophisticated users without an agent access to
> usenet via http protocol of their browser. But it also forces the
> limited functionality of their agent on all their google group users.
I don't follow. App X allows people to do Y, but it also forces upon said
people the limited functionality of X toward Y? I don't think 'allows' and
'forces' go well together in this context.
> They also never retire old and very old items from their database
> which, as Joachim mentions, often results in users responding
> to ancient articles.
How so? It's the reverse, they *disallow* users from replying to old
messages, something you can do with other agents (can't see what's wrong
with that, either). And yes, they do archive everything - that's their job
and you make it sound like it's wrong.
António Marques
> Peter T. Daniels wrote:
>> On Dec 15, 11:52 pm, "PaulJK"<paul.kr...@paradise.net.nz> wrote:
>>> Peter T. Daniels wrote:
I did actually mean 'ascribe' because I don't think google actually assigns
a permanent number to messages. It's just something it creates on the fly
while displaying threads to enhance legibility/navigation. Of course, that
does not mean 'ascribe' can convey the idea more correctly.
PaulJK
Of course they serve articles in response to your request.
> they only serve web pages which happen to be usable as a GUI to
> access the content of articles.
For Chrissake, what you see is the GUI web page interface
to their internal agent containing their internal database.
Akin to OE agent and its database on my own computer.
Their database is fed articles received by their NNTP server.
> What's next, calling a webmail interface a mail server?
I am not talking about what you see which just a front end
interface to serve web pages to the end user's dumb browser.
I talk about the Google's functionality behind it.
> A mail server is
> something which implements POP and/or IMAP and/or SMTP.
And how do you think they receive and maintain messages
distributed by Usenet? By some magic? All retyped by hand? :-)
Just because you never talk directly to their 'server'
doesn't mean it's not there. You never see past the very
front (html) interface to their version of 'agent', of course
you don't see what's going on further inside between
'agent' and 'server'.
>> What do you think the word 'server' means in this context.
>
> An NNTP server, which is what you were comparing google to, can only mean
> one thing: a server implementing the NNTP protocol. I read your paragraph as
> implying that google did provide some quasi-NNTP interface.
Read it again then!
I implied no such thing. They don't provide an NNTP interface
if any kind to the end user. For umpteenth time, all you you
have is a very restricted idiosyncratic and buggy html interface
to the front end of an 'agent-like' program. You don't see anything
past that front end.
>> That is more or less what zillions of other usenet servers do.
>>
>> What Google does differently is that they put their own idiosyncratic
>> agent in front of it and allow users access to their database
>> through this agent via internet http protocol, specifically html pages.
>
> Not quite; what they do differently is that their NNTP server, if it exists,
> is only accessible internally to their web (server) application. That that
> they provide a web UI to do 'usenet stuff' is a separate thing.
>
>> That allows unsophisticated users without an agent access to
>> usenet via http protocol of their browser. But it also forces the
>> limited functionality of their agent on all their google group users.
>
> I don't follow. App X allows people to do Y, but it also forces upon said
> people the limited functionality of X toward Y? I don't think 'allows' and
> 'forces' go well together in this context.
>
>> They also never retire old and very old items from their database
>> which, as Joachim mentions, often results in users responding
>> to ancient articles.
>
> How so? It's the reverse, they *disallow* users from replying to old
> messages, something you can do with other agents (can't see what's wrong
> with that, either). And yes, they do archive everything - that's their job
> and you make it sound like it's wrong.
I did no such thing. I am not Joachim.
Let's agree that we shall never understand each other,
and call an end to this charade.
PaulJK
You ascribe (credit) parts of play to Shakespeare.
pjk
Joachim Pense
>> How so? It's the reverse, they *disallow* users from replying to old
>> messages, something you can do with other agents (can't see what's wrong
>> with that, either). And yes, they do archive everything - that's their
>> job and you make it sound like it's wrong.
>
> I did no such thing. I am not Joachim.
>
If I made it sound like archiving everything is wrong, then it was
unintentional. Or does your "it" refer to replying to old messages? Well,
replying e.g. to the kremvax thread is sort of vandalism in my view.
Joachim
PaulJK
Sorry, Joachim, I was just vaguely recalling you suggesting that
one shouldn't be able to follow up old posts.
I am not sure if it should be made impossible to do but
I certainly agree with you that it shouldn't be done, unless
one has genuinely good reason for doing so.
pjk
P.S. One problem with the general purpose Google search
is that it returns often undated hits in Usenet groups. A google
user can pick a decades old hit and send a followup to its
group without realizing who the readers are going to be or
how ancient it is.
António Marques
No, they serve web pages.
>> they only serve web pages which happen to be usable as a GUI to
>> access the content of articles.
>
> For Chrissake, what you see is the GUI web page interface
> to their internal agent containing their internal database.
> Akin to OE agent and its database on my own computer.
> Their database is fed articles received by their NNTP server.
What's that got to do with the price of flour?
>> What's next, calling a webmail interface a mail server?
>
> I am not talking about what you see which just a front end
> interface to serve web pages to the end user's dumb browser.
> I talk about the Google's functionality behind it.
But google's functionality behind it, which is not accessible anyway, likely
works just as 'has been done for decades'.
>> A mail server is
>> something which implements POP and/or IMAP and/or SMTP.
>
> And how do you think they receive and maintain messages
> distributed by Usenet? By some magic? All retyped by hand? :-)
>
> Just because you never talk directly to their 'server'
> doesn't mean it's not there. You never see past the very
> front (html) interface to their version of 'agent', of course
> you don't see what's going on further inside between
> 'agent' and 'server'.
Obviously. It is there and it may or may not work just the way NNTP servers do.
>>> What do you think the word 'server' means in this context.
>>
>> An NNTP server, which is what you were comparing google to, can only mean
>> one thing: a server implementing the NNTP protocol. I read your paragraph as
>> implying that google did provide some quasi-NNTP interface.
>
> Read it again then!
> I implied no such thing. They don't provide an NNTP interface
> if any kind to the end user.
You did imply such a thing, by saying:
Every time an agent connects to the (non-Google) server to read a particular
group, it (...) [t]hat is how it has been done for decades, not until Google
introduced its idiotic mindless system that (...)
It did suggest that google introduced some alternative server
protocol/behaviour. At least in an IT context, that's what this kind of
construction is used to.
>>> They also never retire old and very old items from their database
>>> which, as Joachim mentions, often results in users responding
>>> to ancient articles.
>>
>> How so? It's the reverse, they *disallow* users from replying to old
>> messages, something you can do with other agents (can't see what's wrong
>> with that, either). And yes, they do archive everything - that's their job
>> and you make it sound like it's wrong.
>
> I did no such thing. I am not Joachim.
This sounds like sophistry.
> Let's agree that we shall never understand each other,
> and call an end to this charade.
Not until you explain how it is that non-bohemian czechs deal with being
'czechs' in english but not in czech, or how do you tell between something
relating to the bohemian dialect and something relating to the whole language.
António Marques
> PaulJK (in sci.lang):
>
>>> How so? It's the reverse, they *disallow* users from replying to old
>>> messages, something you can do with other agents (can't see what's wrong
>>> with that, either). And yes, they do archive everything - that's their
>>> job and you make it sound like it's wrong.
>>
>> I did no such thing. I am not Joachim.
>>
> If I made it sound like archiving everything is wrong, then it was
> unintentional.
No, you didn't do that. Paul did and doesn't want to admit it.
> Or does your "it" refer to replying to old messages? Well,
> replying e.g. to the kremvax thread is sort of vandalism in my view.
Well, I don't know what the krevmax thread is, but I gave that idea some
consideration, and concluded that it should be up to the vandals to reply to
historical threads. One can always filter out their messages if needed.
António Marques
> A google user can pick a decades old hit and send a followup to its group
> without realizing who the readers are going to be or how ancient it is.
That may or may not be stupid, but in what way is it a problem?
Adam Funk
> P.S. One problem with the general purpose Google search
> is that it returns often undated hits in Usenet groups. A google
> user can pick a decades old hit and send a followup to its
> group without realizing who the readers are going to be or
> how ancient it is.
A related problem with the "Google Groups" search is that AFAICT there
is no way to limit the search to the USENET newsgroups only, unless
you use the group:sci.* option (for example) to limit the search to
one major USENET hierarchy. Google Groups searches indiscriminately
conflate newsgroups, Google's own groups, mailing list archives, and
(I think) even some web fora.
--
I worry that 10 or 15 years from now, [my daughter] will come to me
and say 'Daddy, where were you when they took freedom of the press
away from the Internet?' [Mike Godwin, EFF http://www.eff.org/ ]
António Marques
> António Marques wrote:
>> PaulJK wrote, on 17-12-2009 01:43:
>>> Peter T. Daniels wrote:
>>>> On Dec 15, 11:52 pm, "PaulJK"<paul.kr...@paradise.net.nz> wrote:
>>>>> Peter T. Daniels wrote:
>>>>>> On Dec 15, 8:39 am, António Marques<antonio...@sapo.pt>
>>>>>> wrote:
>>>>>>> Everyone not using google groups is probably using something
>>>>>>> that doesn't ascribe numbers to messages. Don't remeber when
>>>>>>> you used Netscape?-
>>>>>>
>>>>>> counting them.
>>>>>
>>>>> ?
>>>>
>>>> Cardinal vs. ordinal -- what does "ascribe numbers to" mean?
>>>
>>> I guess Antonio meant to say 'assign numbers'.
>>
>> I did actually mean 'ascribe' because I don't think google actually
>> assigns a permanent number to messages. It's just something it creates
>> on the fly while displaying threads to enhance legibility/navigation.
>> Of course, that does not mean 'ascribe' can convey the idea more
>> correctly.
>
> You ascribe (credit) parts of play to Shakespeare.
If you didn't know the relative writing dates for Shakespeare's plays, and
you settled on a putative order, could you say you were ascribing numbers to
them?
Peter T. Daniels
You'd say you were ascribing a sequence to them.
If you had a way of dating them, you could ascribe dates to them.
PaulJK
You don't ascribe numbers, you assign numbers.
However, you can ascribe a sequence of numbers.
pjk
PaulJK
Sorry, not "sequence of numbers", just "sequence" or "order"
of them. I think. :-)
> pjk