NNTP client problem (BSD Sockets)

On 2008-03-20, Barry Margolin <barmar@alum.mit.edu> wrote:
> In article <47e1f278$0$16656$4c368faf@roadrunner.com>,
>  Logan Shaw <lshaw-usenet@austin.rr.com> wrote:
> 
>
> If the bug is in a server and only impacts broken clients, and doesn't
> cause any annoying side effects, I suspect neither will give it very
> high priority.
>

Well, if the result of the defect is that a log file grows larger
unnecessarily, that could be considered an annoying side effect. :-)

Even if not, I think some developers would fix the problem in order to
make their system more robust, as long as it's affordable - does not
take a long time to fix/test.


[I didn't see Logan's original response from my news server for some
reason.  Perhaps they have a bug :-)]

--

On Thu, 20 Mar 2008 22:53:12 +0100, Jim Cochrane wrote:

> On 2008-03-20, Barry Margolin <barmar@alum.mit.edu> wrote: 
> Well, if the result of the defect is that a log file grows larger
> unnecessarily, that could be considered an annoying side effect. :-)
>
> Even if not, I think some developers would fix the problem in order to
> make their system more robust, as long as it's affordable - does not
> take a long time to fix/test.
>
>
> [I didn't see Logan's original response from my news server for some
> reason.  Perhaps they have a bug :-)]

:-)
Well in this particular case, the server will receive
command1 args\r\n
\0command2 args2\r\n
\0...

, and will probably handle the first command and hang on the nul
character. For a line based protocol, there will be some way to eat the
nul, and continue, but probably other (worse) things will break instead.
How forgiving must an implementation be ? should it also eat excess
whitespace?
Just hanging on the file descriptor, and timing out eventually will
probably be the sanest way to deal with unbehaved clients like this one.
(and maybe add them automatically to a IP-ban list, too :-)

HTH,
AvK

In article <slrnfu5n67.3me.allergic-to-spam@no-spam-allowed.org>,
Jim Cochrane <allergic-to-spam@no-spam-allowed.org> wrote:

> On 2008-03-20, Barry Margolin <barmar@alum.mit.edu> wrote: 
>
> Well, if the result of the defect is that a log file grows larger
> unnecessarily, that could be considered an annoying side effect. :-)

If it only happens when some random is programming a toy client
implementation, it's not likely to even be noticed.  If there were a
common client that had the bug that triggered the error, then it might
produce enough volume for them to care.

> Even if not, I think some developers would fix the problem in order to
> make their system more robust, as long as it's affordable - does not
> take a long time to fix/test.

And if they don't have more important things to do.  Like fix bugs that
affect real users, not just novice programmers who don't know how to
call write() properly.

--
Barry Margolin, barmar@alum.mit.edu
Arlington, MA
*** PLEASE don't copy me on replies, I'll read them in the group ***

On 2008-03-20 23:02:52 +0000, Barry Margolin <barmar@alum.mit.edu> said:
>
> And if they don't have more important things to do.  Like fix bugs that
> affect real users, not just novice programmers who don't know how to
> call write() properly.

Nice...

Fancy trying to be any more condecending?

Anyway, as for the thread it has been useful, certainly some things I
had not considered. I've looked into setting up a simple news server so
that I don't need to clog Giganews servers. Although one would have to
question if they really would get annoyed at what, a max of 15 - 20
connection attempts a w?

On Mar 20, 5:19 pm, Niz <n...@nicetry.com> wrote:

> Anyway, as for the thread it has been useful, certainly some things I
> had not considered. I've looked into setting up a simple news server so
> that I don't need to clog Giganews servers. Although one would have to
> question if they really would get annoyed at what, a max of 15 - 20
> connection attempts a w?

Sending zero bytes to a server that expects text can crash the server
if it's poorly written. It's very unlikely that giganews servers have
any bugs that serious, and I personally don't think it's particularly
bad to use a public-facing server to test your code.

There are enough people who explicitly permit such things though --
why not use them?

DS

Niz <niz@nicetry.com> wrote:
<snip>
> Anyway, as for the thread it has been useful, certainly some things I
> had not considered. I've looked into setting up a simple news server so
> that I don't need to clog Giganews servers. Although one would have to
> question if they really would get annoyed at what, a max of 15 - 20
> connection attempts a w?

I'm confident they have no idea you're testing, and that their server
software can handle the sort of data you sent. With the traffic they see,
and just random perturbations from the network, the logs are littered with
junk, anyhow. (Of course, if reports are made from the logs, aborted
sessions like this can be really annoying to weed out.)

I don't think its improper to test your software against, more-or-less
public servers like this (being such a large provider, they're more likely
to be especially resilient). Just that, the privilege shouldn't be abused.
And one way to make sure its not abused is that developers take due
diligence, which includes making sure their software is reasonably
correct.[1]


[1] I realize that actual bugs are likely to be stressed by protocol
features and extensions for which a developer doesn't have much choice but
to test outside his environment. And spammers blunt the efficacy of being
good netizens. But, that kind of attitude never got anybody anywhere. Plus,
when you unload on some network engineer for needlessly blocking ICMP, you
have good moral standing.

In article <2008032100191316807-niz@nicetrycom>, Niz <niz@nicetry.com>
wrote:

> On 2008-03-20 23:02:52 +0000, Barry Margolin <barmar@alum.mit.edu> said: 
>
> Nice...

Sorry, forgot you might still be in this thread.

> Fancy trying to be any more condecending?

Better develop a thick skin if you're gonna post your code to the
Internet. :)

> Anyway, as for the thread it has been useful, certainly some things I
> had not considered. I've looked into setting up a simple news server so
> that I don't need to clog Giganews servers. Although one would have to
> question if they really would get annoyed at what, a max of 15 - 20
> connection attempts a w?

As others have said, unless this is actually causing their server to
crash, which I hope is unlikely, I doubt they'll ever notice.

--
Barry Margolin, barmar@alum.mit.edu
Arlington, MA
*** PLEASE don't copy me on replies, I'll read them in the group ***