From
[email protected] Thu Feb 8 12:19:15 EST 1996
Article: 18041 of news.software.nntp
Path: news.math.psu.edu!news.iag.net!news.worldpath.net!news.cic.net!condor.ic.net!news2.acs.oakland.edu!jobone!newsxfer.itd.umich.edu!news.eecs.umich.edu!usenet
From:
[email protected] (Michael Hucka)
Newsgroups: news.software.nntp,news.software.readers
Subject: Re: TRN/INN Repeatedly asking about same "new" newsgroups
Date: 07 Feb 1996 20:11:33 -0500
Organization: University of Michigan EECS, Ann Arbor, Mich., USA
Lines: 152
Sender:
[email protected]
Distribution: inet
Message-ID: <
[email protected]>
References: <
[email protected]>
Reply-To:
[email protected]
NNTP-Posting-Host: hawk.eecs.umich.edu
In-reply-to:
[email protected]'s message of 7 Feb 1996 18:15:25 -0500
X-Attribution: Mike Hucka
To:
[email protected] (Todd D. Taft)
X-Newsreader: Gnus v5.1
Xref: news.math.psu.edu news.software.nntp:18041 news.software.readers:22491
>>>>> On 7 Feb 1996,
[email protected] (Todd D. Taft) wrote:
taft> Recently our news reader programs (primarily trn) have started
taft> telling our users that new newsgroups have been created, and asking
taft> users if they would like to subscribe to these certain newsgroups
taft> every time that they start up the program. It will only ask about
taft> newsgroups that are not listed in the user's .newsrc file, but it
taft> won't ask about every newsgroup not listed in the .newsrc file. The
taft> only pattern that I can detect is that the ones that users keep
taft> getting prompted about have been newgrouped on our server relatively
taft> recently. Our server used to allow users to list only the newsgroups
taft> .... [rest deleted] ....
This kind of sounds like the trn leap year bug. Attached is the patch
posted last month:
From:
[email protected] ( Anne Bennett )
Subject: Re: trn-3.4.1 keeps asking about new groups!
Date: 11 Jan 1996 20:44:37 GMT
Newsgroups: news.software.readers,news.admin.misc,news.software.nntp
Organization: Concordia University, Montreal, Canada
Lines: 124
Message-ID: <
[email protected]>
References: <
[email protected]> <
[email protected]> <
[email protected]>
I wrote:
> That patch is not correct [...]
> A better patch is being
> elaborated on the trn testers' mailing list. I can post it here when
> it's done.
1) I was wrong; the patch was OK, at least until 2100. I re-append it here
as the "simple patch".
2) A more correct but more complicated patch, appended here as "fancy
patch", was submitted, which will work until 2400.
Wayne Davison, author of trn, writes that he will go with the simple patch
because the resulting calculation is faster, and he hopes no-one will still
be using trn in the year 2100. :-) I am using the fancy patch at my site.
Pick your own favourite; either will work. BTW, you should have no trouble
applying these patches to strn-0.9.2, if that's what you're using, though
the line numbers are for trn-3.6.
Anne.
--
Ms. Anne Bennett
[email protected] (514) 848-7606
Computing Services, Concordia University, Montreal QC, Canada H3G 1M8
(Administrator for Concordia University mail relay and news transfer)
----------------------------------------------------------------------------
SIMPLE PATCH THAT WORKS UNTIL 2100
Submitted by Miquel van Smoorenburg <
[email protected]>, reposted
to trn-test list by
[email protected] (John Line)>
===== the patch (to nntp.c, line numbers are for trn 3.6)
------- nntp.c -------
*** /tmp/da005NK Thu Jan 1 00:00:00 1970
--- nntp.c Wed Jan 3 20:43:11 1996
***************
*** 319,326 ****
for (month--; month; month--)
day += maxdays[month];
! ss = ((((year-1970) * 365 + (year-1968)/4 + day - 1) * 24L + hh) * 60
! + mm) * 60 + ss;
return ss;
}
--- 319,326 ----
for (month--; month; month--)
day += maxdays[month];
! ss = ((((year-1970) * 365 + (year-1969)/4 + day - 1) * 24L + hh) * 60
! + mm) * 60 + ss;
return ss;
}
----------------------------------------------------------------------------
FANCY PATCH THAT WORKS UNTIL 2400
From:
[email protected] (Paul Close)
Subject: Re: trn (3.6 & 4.0-test38) listing "new groups" that aren't new
To:
[email protected] (TRN mailing list)
Date: Fri, 5 Jan 1996 16:43:27 -0800 (PST)
Okay, here's my fix. This uses a proper leap year calculation, and should
work for a very long time :-)
--- trn-3.6/nntp.c Fri Nov 18 22:01:23 1994
+++ nntp.c.leap Fri Jan 5 16:38:33 1996
@@ -285,12 +285,13 @@
/* This is a 1-relative list */
static int maxdays[] = { 0, 31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31 };
+#define isleap(y) ((((y) % 4) == 0 && ((y) % 100) != 0) || ((y) % 400) == 0)
time_t
nntp_time()
{
char *s;
- int year, month, day, hh, mm;
+ int year, month, day, hh, mm, y;
time_t ss;
nntp_command("XDATE" + (CompliantServer? 0 : 1));
@@ -306,11 +307,11 @@
s[4] = '\0';
year = atoi(s);
- /* This simple algorithm will be valid until the year 2400 */
- if (year % 4)
- maxdays[2] = 28;
- else
+ if (isleap(year))
maxdays[2] = 29;
+ else
+ maxdays[2] = 28;
+
if (month < 1 || month > 12 || day < 1 || day > maxdays[month]
|| hh < 0 || hh > 23 || mm < 0 || mm > 59
|| ss < 0 || ss > 59)
@@ -319,8 +320,12 @@
for (month--; month; month--)
day += maxdays[month];
- ss = ((((year-1970) * 365 + (year-1968)/4 + day - 1) * 24L + hh) * 60
- + mm) * 60 + ss;
+ for (y = 1970; y < year; y++)
+ if (isleap(y))
+ day += 366;
+ else
+ day += 365;
+ ss += (((day - 1) * 24L + hh) * 60 + mm) * 60;
return ss;
}
----------------------------------------------------------------------------
--
Mike Hucka University of Michigan Artificial Intelligence Lab
[email protected] (X.500) and EECS Departmental Computing Organization
http://ai.eecs.umich.edu/people/hucka/ Ann Arbor, Michigan 48109, U.S.A.
