wls@astrovax.UUCP (William L. Sebok) (10/15/84)
The following addition to news 2.10.2 has been posted before but it is so
important that I think it needs reiterating. This addition implements a
C pre-processor variable NETPATHS that causes the reply command of readnews
and vnews to use optimized network paths rather than the roundabout news
arrival path. NETPATHS is defined in defs.h to be the name of a dbm-style
file.
This file, uses site names as keys, and under these keys stores the
path to the site with a %s in place of a name of a user at that site. For
example, under the key "sitename" might be stored the string:
site1!site2!sitename!%s
A file of this form can be generated with pathalias.
If a site is not found in the database the path up to the first site in the
news path present in the database is used.
I have a somewhat different philosophy here than Chuq Von Rospach, who has
optimization invoked by a -o option. Optimization is the default. It is the
naive user I want make use the optimized paths. For the sophisticated user
I have provided the News-path: field in the header which contains the
convoluted news path. This sophisticated user can edit the full path back
by hand if he/she is so inclined. I have intentionally made the user have to
go to some extra work to get un-optimized paths.
It is my strong belief that these additions or some other hooks for path
optimizing should be part of the "official distribution". After all it is
an option. With the posting of the Uucp map I do not believe that there is
any excuse for replies with long paths bouncing around the net. If some
hooks for path optimization are not included in the official distribution
there will be no justification for the net powers to complain about the money
wasted by replies passing through large numbers of sites.
Bill Sebok Princeton University, Astrophysics
{allegra,akgua,burl,cbosgd,decvax,ihnp4,noao,princeton,vax135}!astrovax!wls
*** defs.h.ORIG Mon Oct 15 01:19:25 1984
--- defs.h Mon Oct 15 01:18:13 1984
***************
*** 47,52
/* the .newsrc */
#define ZAPNOTES /* if you want old style notes headers moved
/* into the headers */
#define BSD4_2 /* if you are running 4.2 BSD */
/* #define BSD4_1C /* if you are running 4.1c BSD */
#define SENDMAIL "/usr/lib/sendmail -oi -oem" /* command line to run
--- 47,54 -----
/* the .newsrc */
#define ZAPNOTES /* if you want old style notes headers moved
/* into the headers */
+ #define NETPATHS "/usr/lib/uucp/alpath" /* compiles in pathalias optimizing */
+ /* code to read,vnews */
#define BSD4_2 /* if you are running 4.2 BSD */
/* #define BSD4_1C /* if you are running 4.1c BSD */
#define SENDMAIL "/usr/lib/sendmail -oi -oem" /* command line to run
*** visual.c.ORIG Sat Oct 6 20:44:48 1984
--- visual.c Sun Oct 14 23:47:19 1984
***************
*** 17,22
#include <errno.h>
#include "rparams.h"
#if defined(BSD4_2) || defined(BSD4_1C)
#include <sys/dir.h>
#else
--- 17,28 -----
#include <errno.h>
#include "rparams.h"
+ #ifdef NETPATHS
+ # include <dbm.h>
+ # ifdef NULL
+ # undef NULL
+ # endif NULL
+ #endif NETPATHS
#if defined(BSD4_2) || defined(BSD4_1C)
#include <sys/dir.h>
#else
***************
*** 811,816
char *replyname();
struct stat statb;
time_t creatm;
/* Put the user in the editor to create the body of the reply. */
ed = getenv("EDITOR");
--- 839,852 -----
char *replyname();
struct stat statb;
time_t creatm;
+ #ifdef NETPATHS
+ /* stuff for finding paths in pathalias database */
+ static dbmopen = 0;
+ static char newspaths[] = NETPATHS;
+ datum key, result;
+ register char *p1, *p2;
+ char sitename[100];
+ #endif NETPATHS
/* Put the user in the editor to create the body of the reply. */
ed = getenv("EDITOR");
***************
*** 843,851
else
#endif
p = replyname(h);
! fprintf(rfp, "To: %s\n", p);
! fprintf(rfp, "Subject: %s\n", subj);
! fprintf(rfp, "In-reply-to: your article %s\n", h->ident);
sprintf(rcbuf, "%s -t < %s; rm -f %s", MAILPARSER, tf, tf);
nogomsg = "Mail not sent";
putc('\n', rfp);
--- 879,917 -----
else
#endif
p = replyname(h);
! #ifdef NETPATHS
! /* Use data Base to Find a reply a Path to a site */
! if (dbmopen==0)
! dbmopen = (dbminit(newspaths)==0) ? 1 : -1 ;
!
! result.dptr = NULL;
! if (dbmopen>0) {
! for(p1 = p; (p2 = index(p1,'!'))!=NULL; p1 = p2+1);
! if (p != p1) {
! key.dptr = sitename;
! p1--;
! do {
! for (p2=p1; (p1!=p) && (*--p1!='!'); );
! key.dsize = p2 - p1 ;
! strncpy(sitename, p1+1, key.dsize);
! sitename[key.dsize-1] = '\0';
! result = fetch(key);
! } while ( (p1 != p) && result.dptr == NULL);
! }
! }
! if (dbmopen > 0 && result.dptr != NULL) {
! fprintf(rfp, "To: ");
! fprintf(rfp, result.dptr, p2+1);
! fprintf(rfp, "\nSubject: %s\n", subj);
! fprintf(rfp, "In-reply-to: your article %s\n", h->ident);
! fprintf(rfp, "News-path: %s\n", p);
! } else
! #endif NETPATHS
! {
! fprintf(rfp, "To: %s\n", p);
! fprintf(rfp, "Subject: %s\n", subj);
! fprintf(rfp, "In-reply-to: your article %s\n", h->ident);
! }
sprintf(rcbuf, "%s -t < %s; rm -f %s", MAILPARSER, tf, tf);
nogomsg = "Mail not sent";
putc('\n', rfp);
********************
*** readr.c.ORIG Sat Oct 6 20:13:40 1984
--- readr.c Sun Oct 14 23:48:24 1984
***************
*** 7,12
#endif !lint
#include "rparams.h"
#if defined(BSD4_2) || defined(BSD4_1C)
#include <sys/dir.h>
#else
--- 7,20 -----
#endif !lint
#include "rparams.h"
+
+ #ifdef NETPATHS
+ # include <dbm.h>
+ # ifdef NULL
+ # undef NULL
+ # endif NULL
+ #endif NETPATHS
+
#if defined(BSD4_2) || defined(BSD4_1C)
#include <sys/dir.h>
#else
***************
*** 616,621
char folbuf[BUFLEN];
struct stat statb;
long creatm;
hptr = h;
while (*bptr && index("d-", *bptr)) {
--- 644,657 -----
char folbuf[BUFLEN];
struct stat statb;
long creatm;
+ #ifdef NETPATHS
+ /* stuff for finding paths from pathalias database */
+ static int dbmopen = 0;
+ static char newspaths[] = NETPATHS;
+ datum key, result;
+ register char *p1, *p2;
+ char sitename[100];
+ #endif NETPATHS
hptr = h;
***************
*** 649,654
*rcbuf = '\0';
pathptr = replyname(hptr);;
for (ptr1 = address, ptr2 = pathptr; *ptr2; ptr1++, ptr2++) {
if (index("\"\\$", *ptr2))
*ptr1++ = '\\';
--- 686,717 -----
*rcbuf = '\0';
pathptr = replyname(hptr);;
+ #ifdef NETPATHS
+ /* Use data Base to Find a reply a Path to a site --- W. Sebok */
+ if (dbmopen==0)
+ dbmopen = (dbminit(newspaths)==0) ? 1 : -1 ;
+
+ result.dptr = NULL;
+ if (dbmopen>0) {
+ for(p1 = pathptr; (p2 = index(p1,'!')) != NULL; p1 = p2+1);
+ if (pathptr != p1) {
+ key.dptr = sitename;
+ p1--;
+ do {
+ for ( p2=p1; (p1!=pathptr) && (*--p1 != '!'); );
+ key.dsize = p2 - p1 ;
+ strncpy(sitename, p1+1, key.dsize);
+ sitename[key.dsize-1] = '\0';
+ result = fetch(key);
+ } while ( (p1 != pathptr) && result.dptr==NULL);
+ }
+ }
+ if (result.dptr != NULL) {
+ sprintf(sitename, result.dptr, p2+1);
+ p1 = pathptr;
+ pathptr = sitename;
+ }
+ #endif NETPATHS
for (ptr1 = address, ptr2 = pathptr; *ptr2; ptr1++, ptr2++) {
if (index("\"\\$", *ptr2))
*ptr1++ = '\\';
***************
*** 695,700
} else {
fprintf(tfp, "To: %s\n", pathptr);
fprintf(tfp, "Subject: %s\n", subj);
fprintf(tfp, "References: %s\n\n", folbuf);
fstat(fileno(tfp), &statb);
creatm = statb.st_mtime;
--- 758,767 -----
} else {
fprintf(tfp, "To: %s\n", pathptr);
fprintf(tfp, "Subject: %s\n", subj);
+ #ifdef NETPATHS
+ if (dbmopen > 0 && result.dptr != NULL)
+ fprintf(tfp,"News-path: %s\n",p1);
+ #endif NETPATHS
fprintf(tfp, "References: %s\n\n", folbuf);
fstat(fileno(tfp), &statb);
creatm = statb.st_mtime;
************
--
Bill Sebok Princeton University, Astrophysics
{allegra,akgua,burl,cbosgd,decvax,ihnp4,noao,princeton,vax135}!astrovax!wlsmark@cbosgd.UUCP (Mark Horton) (10/15/84)
The official distribution HAS had such a feature since 2.10, over a year ago. You just define INTERNET and replies use the From line rather than the Path line. If you have pathalias on your system, you probably have a mailer that is capable of dealing with internet addresses, or can make it happen trivially. The Path line is not intended for replies, it's intended to keep articles from looping around the network. The fact that it's being used for replies just means that people don't have routing software.
wls@astrovax.UUCP (William L. Sebok) (10/17/84)
> The official distribution HAS had such a feature since 2.10, over > a year ago. You just define INTERNET and replies use the From > line rather than the Path line. > > If you have pathalias on your system, you probably have a mailer > that is capable of dealing with internet addresses, or can make it > happen trivially. > > The Path line is not intended for replies, it's intended to keep > articles from looping around the network. The fact that it's being > used for replies just means that people don't have routing software. I disagree totally. My own feeling is that the person sending the reply should have control of the path and be able to override a (possibly erroneous) smart mailer. I have a smart mailer but don't trust it unless I can see what it is doing. Also the patch inside vnews is able to make use of the additional information in the Path line when the destination site itself is not present in the data base. -- Bill Sebok Princeton University, Astrophysics {allegra,akgua,burl,cbosgd,decvax,ihnp4,noao,princeton,vax135}!astrovax!wls
piet@mcvax.UUCP (Piet Beertema) (10/19/84)
<...> >I have a smart mailer but don't trust it unless I can see >what it is doing. Stick to yesterday's computers that still have all those wonderful lights, displays and maybe loudspeakers that show you what it's doing.... -- Piet Beertema, CWI, Amsterdam ...{decvax,philabs}!mcvax!piet
wls@astrovax.UUCP (William L. Sebok) (10/24/84)
>>I have a smart mailer but don't trust it unless I can see >>what it is doing. > Stick to yesterday's computers that still have all those wonderful lights, > displays and maybe loudspeakers that show you what it's doing.... > -- > Piet Beertema, CWI, Amsterdam > ...{decvax,philabs}!mcvax!piet If one defines INTERNET in the news software even if one knows that the path to the site it is going to use is wrong there is no way to overide it. This is my fundamental objection to the INTERNET option of news. And yes I think that the removal of all lights from computers was a big step backwards. -- Bill Sebok Princeton University, Astrophysics {allegra,akgua,burl,cbosgd,decvax,ihnp4,noao,princeton,vax135}!astrovax!wls
piet@mcvax.UUCP (Piet Beertema) (10/25/84)
<...> >If one defines INTERNET in the news software even if one knows that >the path to the site it is going to use is wrong there is no way to >overide it. This is my fundamental objection to the INTERNET option >of news. Now THAT has nothing to do with the INTERNET option in news, but everything with how intelligent (or not) your mailer is. If you do a reply and are thrown into your favorite editor you can always alter the "To:" line, e.g. using the info from the "Path:" line. And then it depends on your mailer whether or not it does (re)routing regardless of the path you gave. Our (re)routing strategy is not to do (re)routing unless the first site in the given path isn't one of our uucp connections. Thus "user@astrovax" and "astrovax!user" will be routed, but "decvax!.....!astrovax!user" not. -- Piet Beertema, CWI, Amsterdam ...{decvax,philabs}!mcvax!piet
biep@klipper.UUCP (J. A. "Biep" Durieux) (10/29/84)
[] >I have a smart mailer but don't trust it unless I can see >what it is doing. >-- >Bill Sebok Princeton University, Astrophysics >{allegra,akgua,burl,cbosgd,decvax,ihnp4,noao,princeton,vax135}!astrovax!wls If you don't trust some servant, there may be two reasons: either he is stupid and may do things the wrong way, or he is disloyal. If you don't trust a *smart* mailer, then it must be disloyal. Servants tend to turn disloyal by being mistreated by their masters. -- Biep. {seismo|decvax|philabs}!mcvax!vu44!botter!klipper!biep I utterly disagree with everything you are saying, but I am prepared to fight myself to death for your right to say it. --Voltaire
jhull@spp2.UUCP (11/06/84)
> [] > > >I have a smart mailer but don't trust it unless I can see > >what it is doing. > >-- > >Bill Sebok Princeton University, Astrophysics > >{allegra,akgua,burl,cbosgd,decvax,ihnp4,noao,princeton,vax135}!astrovax!wls > > If you don't trust some servant, there may be two reasons: either > he is stupid and may do things the wrong way, or he is disloyal. > If you don't trust a *smart* mailer, then it must be disloyal. > Servants tend to turn disloyal by being mistreated by their masters. > > Biep. > {seismo|decvax|philabs}!mcvax!vu44!botter!klipper!biep This raises a question for Bill: Is the path you are objecting to unworkable or non-optimal? If the former, we all have a problem. If the latter, perhaps we are defining the current state of the art. -- Blessed Be, Jeff Hull trwspp!spp2!jhull@trwrb.UUCP jhull@spp2.UUCP
wls@astrovax.UUCP (William L. Sebok) (11/21/84)
> > >I have a smart mailer but don't trust it unless I can see > > >what it is doing. > > >Bill Sebok Princeton University, Astrophysics > > > > If you don't trust some servant, there may be two reasons: either > > he is stupid and may do things the wrong way, or he is disloyal. > > If you don't trust a *smart* mailer, then it must be disloyal. > > Servants tend to turn disloyal by being mistreated by their masters. > > {seismo|decvax|philabs}!mcvax!vu44!botter!klipper!biep > This raises a question for Bill: Is the path you are objecting to > unworkable or non-optimal? If the former, we all have a problem. If > the latter, perhaps we are defining the current state of the art. > Jeff Hull jhull@spp2.UUCP Both, but the unworkable paths are the main problems. I strongly object to the routing being taken completely out of my hands, so that I can't work around a problem route even if I know that there is a problem. As it takes at least 20 minutes to recompile the routing tables (usually more on this heavily loaded system) it is usually impractical to fix the route that way at right that instant. Someone who isn't site administrator doesn't even have that recourse. -- Bill Sebok Princeton University, Astrophysics {allegra,akgua,burl,cbosgd,decvax,ihnp4,noao,princeton,vax135}!astrovax!wls