Send inn-workers mailing list submissions to
[email protected]
To subscribe or unsubscribe via the World Wide Web, visit
https://lists.isc.org/mailman/listinfo/inn-workers
or, via email, send a message with subject or body 'help' to
[email protected]
You can reach the person managing the list at
[email protected]
When replying, please edit your Subject line so it is more specific
than "Re: Contents of inn-workers digest..."
Today's Topics:
1. Re: QIO_BUFFERSIZE (Julien ?LIE)
2. INN 2.6.4 release candidate (Julien ?LIE)
3. Re: QIO_BUFFERSIZE (Bo Lindbergh)
----------------------------------------------------------------------
Message: 1
Date: Thu, 3 Dec 2020 17:17:24 +0100
From: Julien ?LIE <[email protected]>
To: [email protected]
Subject: Re: QIO_BUFFERSIZE
Message-ID: <[email protected]>
Content-Type: text/plain; charset=utf-8; format=flowed
Hi Bo,
> lib/qio.c contains:
>> /* A reasonable default buffer size to use. */
>> #define QIO_BUFFERSIZE 8192
>
> I don't think this is reasonable any more, because it limits the maximum
> length of an overview line in makehistory, and there are definitely
> artic|es in the wild with longer overview data. Having different overview
> size limits in innd and makehistory is just wrong.
Have you tried to increase that value, and is everything still working
fine? (access to previous records, and write of new ones)
This value is only used for setting the buffer size in QIOfdopen(). It
seems that increasing it would do no harm (contrary to decreasing it,
which then may lead to errors for too long lines).
So we would need to at least have 4 * QIO_BUFFERSIZE because an inferior
value may cause the real buffer size decrease, according to:
if (size > (4 * QIO_BUFFERSIZE) || size == 0)
size = QIO_BUFFERSIZE;
>> $ echo 'group rec.arts.comics.strips
>> over 56677' | nc nntp.aioe.org 119
>> 200 nntp.aioe.org InterNetNews NNRP server INN 2.5.4 ready (posting ok)
>> 211 1224 56506 57729 rec.arts.comics.strips
>> 224 Overview information for 56677 follows
Indeed, that overview line contains about 10179 bytes.
--
Julien ?LIE
??Boire du caf? emp?che de dormir. Par contre, dormir emp?che de boire
du caf?.?? (Philippe Geluck)
------------------------------
Message: 2
Date: Thu, 3 Dec 2020 17:17:49 +0100
From: Julien ?LIE <[email protected]>
To: "[email protected]" <[email protected]>
Subject: INN 2.6.4 release candidate
Message-ID: <[email protected]>
Content-Type: text/plain; charset=utf-8; format=flowed
Hi all,
A release candidate for INN 2.6.4 can be downloaded at:
https://ftp.isc.org/isc/inn/testing/inn-2.6.4rc1.tar.gz
The MD5 checksum is: cfd9ed26b8083f3ad2e9d2023aeae8a9.
A patch from 2.6.3 to 2.6.4rc1 is available in the same directory.
Feel free to report any issue you may encounter.
The final release is scheduled in January.
Known issue (and already fixed in daily snapshots): if the build fails
with an error like "ld: unknown option: -z", just edit Makefile.global
and change the following line:
LDFLAGS = -pie -Wl,-z,relro -Wl,-z,now
to:
LDFLAGS = -pie
and re-run "make".
Changes from 2.6.3 to 2.6.4
* nnrpd now adapts the length of the DH parameter used during a DHE key
exchange so as to comply with the security level OpenSSL 1.1.0 or
later expects. Thanks to Michael Baeuerle for the bug report.
* cnfsstat now also returns information about retired CNFS buffers
(mentioned in cycbuff.conf as a cycbuff but not in a metacycbuff).
* Switch default innreport behaviour to common practice of
externalizing
CSS into a separate file. Its name can be configured with the
*html_css_url* parameter in innreport.conf. If this parameter is
unset, the default innreport.css file name will be used and innreport
will generate this CSS file for you. Previously generated
reports are
kept untouched, though, and will still contain inline CSS if you had
not already set the *html_css_url* parameter in previous INN
versions.
Thanks to Richard Kettlewell for the patch.
* sm can now read and store any number of articles given in wire format
on its standard input when both -s and -R are used. Only native
format was previously possible. Thanks to Bo Lindbergh for the
patch.
* Added new -a flag to rnews to disallow, if needed, the use of
additional unpackers from the rnews.libexec sub-directory of
*pathbin*
(as set in inn.conf); only "rnews" and "cunbatch" will then be
recognized as valid batch commands.
* Added new -b flag to rnews to save rejected articles in the bad
sub-directory of *pathincoming* (as set in inn.conf). Otherwise,
rnews just logs and discards any articles that are rejected or cannot
be parsed for some reason.
* Added new -d flag to rnews to log via syslog the message-ID and the
Path header value of each article rejected as a duplicate.
* Added new --enable-hardening-flags configure-time option, enabled by
default, to use hardening build flags like "-fPIE" and
"-fstack-protector-strong". This option can easily be disabled
if the
compiler or the platform does not support them well. More hardening
build flags will eventually be added in future releases.
--
Julien ?LIE
??Boire du caf? emp?che de dormir. Par contre, dormir emp?che de boire
du caf?.?? (Philippe Geluck)
------------------------------
Message: 3
Date: Thu, 3 Dec 2020 18:11:30 +0100
From: Bo Lindbergh <[email protected]>
To: [email protected]
Subject: Re: QIO_BUFFERSIZE
Message-ID: <[email protected]>
Content-Type: text/plain; charset=us-ascii
Quoth <[email protected]>:
> Have you tried to increase that value, and is everything still working fine?
> (access to previous records, and write of new ones)
Doubling the value makes the "overview line %d is too long" warnings
from makehistory go away. No obvious negative effects, but I haven't
gone looking for non-obvious ones.
/Bo Lindbergh
------------------------------
Subject: Digest Footer
_______________________________________________
inn-workers mailing list
[email protected]
https://lists.isc.org/mailman/listinfo/inn-workers
------------------------------
End of inn-workers Digest, Vol 126, Issue 3
*******************************************
