summaryrefslogtreecommitdiff
path: root/info/digests/texline/no11/nelson.tex
blob: a7fea8fcc76ae6f016512af6824148d9a8a92553 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
\let\\\cr                                        
\def\CMD#1{{\tt #1}}
\def\FILE#1{{\tt #1}}
\def\FTP{{\tt ftp}}
\def\HOSTNAME#1{{\def\-{\discretionary{}{}{}}\tt #1}}
\def\NETLIB{{\tt netlib}}
\def\TUGLIB{{\tt tuglib}}
\let\em\it 
\centerline{\bf The \TUGLIB{} Server}
\smallskip
\noindent 
%\author{Nelson H. F. Beebe}
%\address{Center for Scientific Computing\\
%        Department of Mathematics\\
%        South Physics Building\\
%        University of Utah\\
%        Salt Lake City, UT 84112\\
%        USA\\
%        FAX: (801) 581-4148\\
%        Tel: (801) 581-5254}
% 
%\netaddress[\network{Internet}]{Beebe@science.utah.edu}
%\begin{document}
Scores of sites on the worldwide Internet now
provide access to assorted collections of software
relating to \TeX{} and \MF{}.  In many cases,
these are accessible only via the Internet
mechanism known as {\em anonymous \FTP{}}, a scheme
that permits logins from unknown users, usually on
other machines, with very restricted access.  The
name \FTP{} is an acronym for {\em file
transfer protocol}.
 
Generally, only a portion of the file tree is
visible to the anonymous user, and the command
repertoire is usually limited to little more than
directory listings and file retrieval.  Only a few
sites permit the anonymous user to deposit files
in the anonymous login directory.  Anonymous \FTP{}
provides a means whereby individual remote users
can access file archives, browse around in the
file tree, and retrieve selected files, all
without troubling the staff or other users of the
local machine.
 
While anonymous \FTP{} has been enormously useful to
the Internet community, it is available only
between sites that have direct Internet
connections, and on one of which, anonymous \FTP{}
logins have been enabled.  Sites with only
electronic mail connections to the Internet, such
as those on other networks, like Bitnet, Junet,
SPAN, and Usenet, and those on networks which are
incompatible with the Internet, such as \Janet\ in
the UK, are prevented from using anonymous \FTP{}.
Similarly, sites on the Internet that have
security restrictions, which includes many
commercial, government, and military connections,
may have restrictions that allow only {\it email{}}
access.
 
\section{The \TUGLIB{} connection}
To improve the access to the \TeX{} archives and
other software at Utah, I have installed a
modified version of the \NETLIB{} server
described by Dongarra, which has been renamed
\TUGLIB{}.  This server provides a means whereby
remote users can send electronic mail messages
containing service requests to a daemon program.
The daemon parses the requests, logs them, and
responds to them.
 
The \TUGLIB{} daemon program runs on a local \unix{}
system at Utah, but the mail access is actually
through a mail forwarding address on another
machine, \HOSTNAME{tuglib@\-science.\-utah.\-edu}.
The reasons for this separation are:
    \item{$\bullet$}
          \HOSTNAME{science.\-utah.\-edu} is a
          more widely-known host with a name which
          has been registered on the Internet for
          several years.  It is therefore likely
          to be known on those machines which
          still have not upgraded from fixed host
          tables to domain name servers for
          Internet addressing.
    \item{$\bullet$}
          The \TUGLIB{} software runs only on the
          \unix{} operating system.
          \HOSTNAME{science.\-utah.\-edu} is a
          DEC-20/60 running TOPS-20, but in late
          1990, it will likely be retired and
          replaced by a \unix{} system that will
          answer to the same Internet host name
          (but a different numeric address).
    \item{$\bullet$}
          Separation of the server from the mail
          drop provides flexibility in
          configuration.  In response to load
          patterns, we could change the machine
          running the \TUGLIB{} daemon without
          having to make the change known to
          thousands of users who might wish to use
          the \TUGLIB{} service.
    \item{$\bullet$}
          Through the wonders of NFS ({\em Network
          File System\/}), the \unix{} system running
          the \TUGLIB{} daemon is able to mount
          the TOPS-20 file system, because
          \HOSTNAME{science.\-utah.\-edu} runs an
          implementation of NFS developed by Mark
          Lottor at SRI.  This makes the archives
          of two quite different machines
          available through a single service.
 
   \item{}\indent       Had we purchased NFS support software
          for VAX VMS, it would have been possible
          to provide access to our VAX 8600 as
          well; that will regrettably not happen,
          because our plans are to retire it a few
          months after the DEC-20.
  
\smallskip\noindent 
The present configuration of \TUGLIB{} provides
several services:
    \item{$\bullet$}
          ask for help;
    \item{$\bullet$}
          list contents of a file directory;
    \item{$\bullet$}
          find a file in the archive;
    \item{$\bullet$}
          send one or more files; binary files are
          automatically encoded into a subset of
          the printable {\sc iso\slash ascii} characters for
          {\it email{}} transmission;
    \item{$\bullet$}
          query the TUG address file for
          membership information;
    \item {$\bullet$}
          check load libraries to determine file
          dependencies, so that if a request is
          made for a particular file, all other
          files that it references are
          automatically sent as well.
\smallskip\noindent
The last capability is not currently used by
\TUGLIB{}, since the bulk of the software in the
distribution is \TeX{} files (macros and fonts),
rather than source code of mathematical libraries
like {\sc linpack} and {\sc eispack} for which \NETLIB{}
library support was originally developed.
 
\TUGLIB{} also provides some internal services,
such as logging of requests (both successful and
failed), and exclusion of users listed on an
`enemies' list.  The latter has not yet been
needed, but support is already there should it
ever become necessary.
 
The log of successes provides a useful record of
utilization of the service that may be needed to
convince local administrators of its value.
 
The log of failures is useful in guarding against
break-in attempts, or hogging of resources.
 
The log can also be for finding out whether
alternate query syntaxes might be useful; for
example,
\CMD{envoyer},
\CMD{get},
\CMD{mail},
\CMD{request},
and
\CMD{sned}
are all recognized as synonyms for the \CMD{send}
command.
 
\TUGLIB{}'s {\it email{}} responses come mostly from
external files, rather than from text embedded in
its programs, making it easy to customize for
particular applications, and updates of textual
information can be installed without recompilation
of the software.
 
\section{Getting help}
The simplest command recognized by \TUGLIB{} is
\CMD{help}.  It produces a response containing a
short description of the \TUGLIB{} service with
sample commands.
 
Synonyms for \CMD{help} include \CMD{directory},
\CMD{index}, \CMD{info}, and \CMD{information}.
 
\section{File retrieval}
Files can be retrieved by commands of the form
\CMD{send filename} or \CMD{send filename from
directory}.  Punctuation is ignored, and
unrecognized words are discarded.  A request like
%
\begintt
Please send me the index from the ftp
directory.  Thanks for your help.
\endtt
%
\noindent
is recognized: you will get both an index, and a
help response.
 
 
Directories are named as in \unix{}, that is, a
series of one or more names, separated by slashes,
with no embedded blanks.  If no directory is
specified, the top-level \TUGLIB{} directory is
assumed.  Here are some examples:
%
\begintt
send index
send index from ftp
send plain.tex from tex/inputs
send uudecode.c from support
send 00tdir.lst from tex/pub/cweb
\endtt
%
\noindent
\unix{} symbolic links (duplicate names for the same
file) are used to make particular file trees
accessible to \TUGLIB{}; the file in the last
example,
\FILE{tex\slash{}pub\slash{}cweb\slash{}00tdir.\-lst},
actually resides elsewhere in the file system (in
fact, on the DEC-20 as the file
\FILE{aps:\-<tex.\-pub>\-00tdir.\-lst}), but
appears to \TUGLIB{} to be the \unix{} file with the
absolute path
\FILE{/tuglib\slash{}tex\slash{}pub\slash{}cweb%
\slash{}00tdir.\-lst}.
 
The top-level \TUGLIB{} directory, \FILE{/tuglib},
contains only a small number of files at present:
%
\medskip
\centerline{\vbox{%
\halign{#\strut\hfil\quad&\vtop{\hsize5cm\noindent\tolerance10000#\strut}\cr
\noalign{\hrule}
\FILE{ftp}       & symbolic link to the anonymous
                   \FTP{} directory on
                   \HOSTNAME{science}\cr
\FILE{support}   & support software, mostly for
                   encoding of binary files into
                   printable characters\cr
\FILE{tex}       & symbolic link to the \TeX{}
                   tree on  \HOSTNAME{science}\cr
\noalign{\hrule}}  }}
\medskip
%
For security reasons, you cannot trick \TUGLIB{}
into sending an arbitrary file from elsewhere in
the file system by specifying an absolute
directory path in the \CMD{send} request.  The
leading slash is stripped, so that the file name
{\em always} appears at or below the \TUGLIB{}
home directory.
 
\TUGLIB{} does not provide any mechanism for
wildcard matching of file names, mostly out of
concern for security, and limiting {\it email{}} traffic.
Only those files that are explicitly listed in
index files are visible via \TUGLIB{} (unless the
remote user is rather good at guessing names).
 
Ideally, each directory accessible to \TUGLIB{}
should have an index file named (naturally),
\FILE{index}, containing names of files and a
short description of their contents.  Here is a
portion of the index from the \FILE{ftp}
directory, slightly edited to fit in these narrow
columns.
 
{\let\tt\smalltt\baselineskip10pt
\begintt
PS:<ANONYMOUS>INDEX..7,  9-Dec-89 17:17:54,
Edit by BEEBE
 
This is an index of files available in the
anonymous ftp login directory on
science.utah.edu.
 
00DIR.CMD           -- FTP command file to
                       retrieve files
                       alphabetically
00DIR.LST           -- alphabetical directory
                       listing
00NEWS.TXT          -- brief announcements of
                       new additions to the
                       collections
00PCDOS.TXT         -- setting up TeX for
                       PC DOS
...
TEX-FOR-APPLE-MACINTOSH.TXT
                    -- sources of Macintosh TeX
TEX-FOR-ARABIC.TXT  -- sources of TeX for
                       Arabic typesetting
TEX-FOR-IBM-PC.TXT  -- sources of IBM PC TeX
TEX-FOR-PICTURES.TXT-- combining graphics with
                       TeX
...
\endtt 
}\noindent
However, keeping such an index up-to-date is a
demanding task, considering that at present, there
are nearly 130 file directories and 8000
files in the \TeX{} tree alone.  Consequently, in
most cases, only major directories will have an
\FILE{index} file.  To supplement these index
files, a batch job is run periodically to
automatically create four special files in every
directory accessible to \TUGLIB{}:
%
\medskip
\centerline{\vbox{%
\halign{#\strut\hfil\quad&\vtop{\hsize6cm\tolerance1000\noindent#\strut}\cr
\noalign{\hrule}
\FILE{00dir.cmd}  & alphabetical list of files as
                    \FTP{} and \TUGLIB{} \CMD{get}
                    commands\\
\FILE{00dir.lst}  & alphabetical verbose directory
                    listing\\
\FILE{00tdir.cmd} & reverse time-ordered list of
                    files as \FTP{} and \TUGLIB{}
                    \CMD{get} commands\\
\FILE{00tdir.lst} & reverse time-ordered verbose
                    directory listing\\
\noalign{\hrule}   }}
}
\smallskip

The \FILE{00dir.\-cmd} and \FILE{00tdir.\-cmd}
files are handy for initiating a retrieval of a
complete file directory, since their contents can
be shipped back almost verbatim as requests to
\TUGLIB{}.  The verbose directory listings contain
file sizes in bytes and the last-write time stamp.
A reverse time-ordered listing makes it easy to
find out what is new.
 
The format of the directory listings depends on
the particular operating system; here is part of
the \FILE{00tdir.\-lst} file in the \FILE{tex}
directory, which resides on TOPS-20; some
reformatting has been necessary to make it fit
here:

{\let\tt\smalltt\baselineskip10pt
\begintt
  APS:<TEX>
00LAST30DAYS.TXT.20;P777752 48 121566(7)
                  23-Aug-90 03:08:41 OPERATOR
00LAST7DAYS.TXT.23;P777752   2 3641(7)
                  23-Aug-90 03:04:47 OPERATOR
00RECENT.LOG.1;P777752       6 2707(36)
                  23-Aug-90 03:02:14 OPERATOR
00DIR.CMD.1;P777752          1 2171(7)
                  21-Aug-90 07:21:07 OPERATOR
00DIR.LST.1;P777752          5 11116(7)
                  21-Aug-90 07:20:59 OPERATOR
00INVERTED-INDEX.TXT.15;P777752   6 15360(7)
                  19-Aug-90 18:42:54 OPERATOR
...
\endtt   }
\noindent
Here is how to dissect one of these entries:
%
\smallskip
\noindent
\halign{\strut\tt#\quad\hfil&#\hfil\cr
\noalign{\hrule}
\FILE{00LAST30DAYS.TXT} & file name \\
\FILE{.15}                  & generation number\\
\FILE{;P777752}             & protection bits\\
\FILE{6}                    & count of disk pages\\
                            &  (512 36-bit words)\\
\FILE{15360}                & count of bytes\\
\FILE{(7)}                  & byte size\\
\FILE{19-Aug-90 18:42:54}   & time of last write\\
\FILE{OPERATOR}             & user who last wrote
                              the file\\
\noalign{\hrule}
}
\smallskip

\noindent
Unlike the \unix{} file system, the TOPS-20 file
system is case-{\em insensitive\/}; file names are
conventionally spelled in upper-case, but you can
write them in lower-case, or even mixed-case.
Text files are normally stored with 7-bit bytes,
which is sufficient for the {\sc ascii} character set;
\TeX{} binary files, and files intended for use on
other systems, have 8-bit bytes; native binary
files have 36-bit bytes.  You can always omit the
generation number, since the default is to return
the highest existing generation.  We could fetch
the sample file by a request of the form \CMD{send
00last30days.txt from tex}.
 
While it would be possible to generate these
directory listing files from the \unix{} host, doing
so would lose the byte-size information, which is
needed for correct \FTP{} access, so I prefer to
generate them on the TOPS-20 host instead.
 
\section{Finding files}
With the large number of directories, and the
limited directory listing access provided by
\TUGLIB{}, finding your way around a file tree as
big as the \TeX{} one can be a daunting task.  To
that end, batch jobs that run at regular intervals
produce other helpful indexes:
%
\smallskip
\centerline{\let\tt\smalltt\vbox{%
\halign{\strut\smalltt#\hfil\quad&\vtop{\hsize4cm\tolerance 
10000\raggedright\noindent#\strut}\cr
\noalign{\hrule}
\FILE{00inverted-index.txt} & inverted index:                             
 for each
                              unique file name in
                              the tree, gives a list of
                              directories that
                              contain it \\
\FILE{00last30days.txt}     & a verbose directory
                              listing of files
                              changed in the last
                              30 days anywhere in
                              the \TeX{} tree\\
\FILE{00last7days.txt}      & a verbose directory
                              listing of files
                              changed in the last
                              7 days anywhere in
                              the \TeX{} tree\\
\noalign{\hrule}}                        }}
%
\smallskip\noindent
Some directories will contain a file which is named
\FILE{00readme.txt}: this gives an overview of the
directory contents.  The \FILE{00} prefixes on
these files are to make them come near the
beginning of directory listings, where they are
more likely to be noticed.  On a case-sensitive
file system like \unix{}, they would probably be
named entirely in upper-case letters, which, in
{\sc ascii}, collate before the lower-case letters
normally used for file names.
 
The large number of files available in an archive
such as the one at Utah makes it rather difficult
for external users to find desired files, since
they are not able to login directly and use
directory listing commands.  The \TUGLIB{}
\CMD{find} command helps to remedy this problem.
It searches two standard files, one of which is
prepared by the \TUGLIB{} maintainer and contains
one-line summaries of the contents of every file
directory, and the other is the
\FILE{00inverted-index.\-txt} file described
above.
 
A request like \CMD{find latex.tex} will produce a
response with the names of all the file
directories that contain the file
\FILE{latex.tex}.  Similarly, the request
\CMD{find music} will list not only the name of
the directory containing music fonts, but also all
of the files found in that directory, because
every line in \FILE{00inverted-index.\-txt}
containing the text `music' is matched.
 
\section{Large files and binary files}
Compared to \FTP{}, electronic mail places some
severe restrictions on file transfers:

  \item{$\bullet$}
        Message lengths are limited.  32 kilobytes
        is a reasonable upper bound; larger
        messages may be delayed, or returned to
        the sender, by some mail gateways.
  \item{$\bullet$}
        Messages may contain only printable
        characters; binary files cannot be sent
        without further encoding.
  \item{$\bullet$}
        Some IBM mainframe mail gateways corrupt
        mail that passes through them by having
        inconsistent inbound and outbound
        translations between {\sc ascii} and {\sc ebcdic}        
character
sets.  These may result in
        irreparable many-to-one mappings;  curly braces are often mapped
        into the letters E and L, which is a
        disaster for \TeX{} and C files.
  \item{$\bullet$}
        File names and file attributes cannot be
        automatically attached to {\it email{}} messages.
  \item{$\bullet$}
        \FTP{} is based on reliable network protocols
        like TCP/IP, and \FTP{} transfers are
        always between pairs of machines, with no
        intermediaries, so file transfers do not
        corrupt files.  {\it Email{}} often goes through
        intermediate machines that alter
        characters, truncate long messages, trim
        long lines or trailing blanks, or just
        discard the message altogether.
 
\item{}\indent
        These problems do not exist for
        Internet-to-Internet electronic mail,
        since it too is based on point-to-point
        reliable protocols, but they often occur
        in {\it email{}} between an Internet site and one
        on some other network, like Bitnet or
        Usenet.
   \item {$\bullet$}
        A line beginning with a period terminates
        the mail message on some systems.

\smallskip\noindent 
To deal with the message length limitation,
\TUGLIB{} automatically splits large files into
parts which are mailed separately with
distinctive headers, like {\tt latex.tex (3 of
18)}.
 
File splitting is desirable even for \FTP{} access,
because long transfers may suffer timeouts that
terminate the connection.  A simple utility,
\FILE{bsplit.\-c}, is available in the
\FILE{support} directory, for splitting binary
files into smaller parts.  To avoid destructive
padding with garbage characters on record-oriented
file systems like VAX VMS, the sizes of each part
(except possibly the last) are chosen to be a
multiple of common file system block sizes,
typically 512 bytes.  For example, the command
%
\begintt
bsplit -32768 fonts.tar
\endtt
%
\noindent
would split the file into 32\,KB parts named
\FILE{fonts.\-tar-001}, \FILE{fonts.\-tar-002},
and so on.  The original file can be recovered by
appending the pieces in order; on a reasonable
file system that provides alphabetically-sorted
directories (or at least the illusion thereof)
this can often by done by a single command using
wildcard pattern matching, as in \unix{}:
%
\begintt
cat fonts.tar-??? >fonts.tar
\endtt
%
\noindent
Users on deficient file systems, like that of \PCDOS, 
may have to work a little harder to
reconstruct the original file.
 
\TUGLIB{} normally refuses to mail very large
files; this limitation is removed by making such
files available in split parts.
 
Binary files are automatically recognized by
\TUGLIB{}, and are sent as {\em xxencoded\/}
files.  A file is regarded as `binary' if it
contains any non-printable characters other than
carriage return or line feed, or if it has lines
longer than 72 characters.
 
Because xxencoding is a new scheme developed for
\TUGLIB{}, a header is appended to the response to
describe the encoding in sufficient detail to
allow the recipient to write a program to decode
the message.  Of course, decoding programs are
available from the \TUGLIB{} \FILE{support}
directory, so in practice, few \TUGLIB{} users
will ever have to write their own decoder.
 
Xxencoding is a generalization of \unix{} uuencoding,
and the \FILE{xxencode} and \FILE{xxdecode}
programs will handle uuencoding as well.
Xxencoding was developed to deal with {\it email{}}
corruption, and to facilitate reassembly of large
messages that have been sent in parts.  Both
encoding schemes represent three 8-bit bytes as
four 6-bit bytes, and output is assembled into
lines less than 65 characters in length, so as to
avoid destructive truncation of long lines by
anti-social mailing software.
 
Uuencoding biases the 6-bit bytes by 32, to move
them into the range of printable {\sc ascii} characters.
One variant of \FILE{uuencode} remaps the encoded
blank ({\sc ascii} 32) to back accent ({\sc ascii} 96), which
is the same character to \FILE{uudecode} (it only
looks at the lower six bits in a character).  This
change removes blanks from the output encoding,
and avoids damage from blank-trimming mailers.
 
Xxencoding instead maps a 6-bit byte into plus,
minus, digits, upper-case letters, or lower-case
letters, which is a 64-character set that is more
likely to be immune to translation corruption.
The translation table is included in the output,
and will be used by \FILE{xxdecode}, so the
encoding can survive one-to-one character
remappings.  \FILE{xxencode} also prefixes each
line with a two-character sequence, `xX', and on
completion of encoding, appends a byte count and a
CRC-16 checksum to the output.
 
Cyclic redundancy checksums are superior to simple
checksums obtained by adding or exclusive-OR'ing
data byte sequences.  Such methods cannot detect
bytes out of sequence, and can fail to detect even
two single-bit errors, such as in two consecutive
bytes with an inverted bit in the same position.
 
By contrast, the CRC-CCITT checksum used by the
ANSI X.25, ADCCP, HDLC, and IBM SDLC protocols
detects error bursts up to 16 bits in length, and
99 percent of error bursts greater than 12 bits.
The CRC-16 checksum used by DDCMP and Bisync, and
by \FILE{xxencode} and \FILE{xxdecode}, detects
error bursts up to 16 bits, and 99 percent of
bursts greater than 16 bits in length.
 
\FILE{xxdecode} ignores any line without the `xX'
prefix, allowing input to consist of a
concatenation of several mail messages without the
necessity of stripping mail headers and trailers.
\FILE{xxdecode} also validates the byte count and
checksum so as to detect corruption. Regrettably,
\FILE{uudecode} has no comparable facility; it
will happily produce garbage from corrupted input
with no warning to the user.
 
Differences in end-of-line terminators on various
operating systems are a minor nuisance; carriage
return (Apple Macintosh), line feed (\unix{}), and
carriage return followed by linefeed (TOPS-20 and
\PCDOS) are all in use.  All of these, and more,
are supported on VAX VMS.  As long as files can be
transferred as plain text, \FTP{} and {\it email{}}
handle `lines', and line terminators appropriate
to the receiving system will automatically be
supplied.  When files are encoded however, the
line terminators are encoded with them.  Thus,
transfer of a file from one of the directories
residing on TOPS-20 to a \unix{} system will result
in a non-native carriage return at the end of
each line.
 
To deal with most of this problem, two utilities,
\FILE{dos2ux} and \FILE{ux2dos}, are provided in a
single shell bundle, \FILE{dos2ux.\-shar}, in the
\FILE{support} directory.  They take a list of
files on the command line, and convert |<cr>| |<lf>| to |<lf>|
or |<lf>| to |<cr>| |<lf>|, and also preserve the last write
date of the original file.  I've not yet written
the \FILE{mac2ux} and \FILE{ux2mac} variants to
handle conversion between |<cr>| and |<lf>| terminators,
but they could be easily generated from the
\FILE{dos2ux} and \FILE{ux2dos} code.  Since the
operations are so similar, it would probably make
sense to merge them into a single utility whose
operation was controlled by a command line option,
or by the name of the file it was stored in.
 
Retrieval of a complete directory having many
small files is painful because of the many
\TUGLIB{} requests needed.  The solution is to
make directory contents available in a single
archive file, such as the \FILE{.arc} format
widely used on \PCDOS, or the \unix{} \FILE{.tar} and
compressed \FILE{.tar.Z} formats.  Besides
allowing a group of files to be retrieved in one
request, the archive file preserves exact file
names and importantly, file time stamps.  We have
made several large collections, including all of
our public fonts, available this way.
 
Public-domain implementations of the \FILE{arc},
\FILE{compress}, and \FILE{tar} utilities are
available for several operating systems, including
\PCDOS, TOPS-20, \unix{}, and VAX VMS, so the use of
these archive formats should not pose a problem
for most \TUGLIB{} clients.
 
\section{TUG membership query}
The TUG address data base is kept in a
specially-formatted secret file in the
\FILE{/tuglib} tree, inaccessible to anyone but
the super-user (\unix{} \FILE{root} login) or the
\TUGLIB{} daemon.  A multi-line address like
%
{\def\\{}\obeylines
Nelson H. F. Beebe\\
Center for Scientific Computing\\
Department of Mathematics\\
220 South Physics Building\\
University of Utah\\
Salt Lake City, UT 84112\\
Tel: (801) 581-5254\\
FAX: (801) 581-4148\\
{\it email{}}: Internet: beebe@science.utah.edu\\
TUG Board of Directors
\smallskip}

%
\noindent
is reformatted into a single-line entry with
unprintable control characters separating the
original lines.
 
The \CMD{whois} (or \CMD{who is}) command uses a
shell script to invoke the \unix{} \FILE{grep}
command to find matching lines in the address file
(ignoring letter case and punctuation), and then
converts the magic separator characters back into
normal newline characters.  Thus, the above entry
could be retrieved by any of several different
commands:
%
\begintt
whois nelson beebe
whois beebe
whois 581-5254
whois SALT laKe CiTy
\endtt
%
\noindent%
Each word in the \CMD{whois} query is matched
separately against the {\em entire\/} address
entry.  You need not remember people's initials,
and you can use \CMD{whois} on parts of the
address other than the personal name.
 
Of course, \TUGLIB{} is still just a stupid
computer program: if you send \CMD{who is Dave
Kellerman}, it will not understand that \CMD{Dave}
is short for \CMD{David} and it will fail to match
\CMD{David Kellerman}.  Personal names can be
abbreviated to a leading prefix, however:
\CMD{whois Don Knuth} will find Donald E.~Knuth's
address.
 
It is quite possible that multiple addresses match
a \CMD{whois} query: \CMD{whois sweden}
potentially can list the addresses of all TUG
members in Sweden.  However, organizations,
including TUG, guard their membership lists with
care, partly because such lists have commercial
value, and partly out of concern for privacy.
Germany, for example, has laws that severely
restrict the use of address data bases.
Consequently, \TUGLIB{} will refuse to send more
than a small number of addresses in response to a
\CMD{whois} command, and it makes no provision for
restarting a \CMD{whois} search.  You cannot
retrieve the entire list by a command like
\CMD{whois *} (for the \unix{} \FILE{grep} utility,
the \CMD{*} matches zero or more of anything, that
is, everything).
 
My original intent in setting up \TUGLIB{} was to
augment it with a mail forwarding service, in order that electronic mail sent to
an address such as \HOSTNAME{malcolm.\-clark@\-science.\-utah.\-edu}
would automatically be mapped to that member's
real Internet address and forwarded.  Such a
forwarding list is maintained for the numerical
analysis community on the machine
\HOSTNAME{na-net.\-stanford.\-edu}; mail to
\HOSTNAME{moler@\-na-net.\-stanford.\-edu} will
always get to Cleve Moler, no matter where he is.
This is a very convenient service, since a
community of researchers can easily keep in touch,
even though they may be moving often.
 
Consultation with the maintainers of
\HOSTNAME{na-net} revealed that a surprising load
is caused by this forwarding service, and at
times, several CPUs are kept busy just handling
the mail forwarding.  Since the machines on which
\TUGLIB{} currently runs have many other duties as
well, and provide \TUGLIB{} as a free community
service in their spare time, I abandoned further
ideas for automatic mail forwarding.
 
The \CMD{whois} command eliminates mail forwarding
overhead, since presumably an address will be
looked up once, and then {\it email{}} correspondence
will be initiated directly by the users
themselves.  \CMD{whois} provides the additional
service of supplying postal addresses, telephone
numbers, TUG committee affiliations, and any other
relevant information that happens to be recorded
in the complete address entries.
 
There is nothing magic about the address data base
handling.  If you wanted to, you could easily
eliminate the restriction on the number of matches
returned, and then implement a recipe data base
lookup to answer queries like \CMD{who is garlic}.
A code change to accept the alternate request form
\CMD{what has garlic} would then be desirable.
 
\section{Future extensions}
Since it built upon the lessons of \NETLIB{}, I
feel confident that \TUGLIB{} is quite
satisfactory in its current configuration.  There
are a few things that I would like to add, if time
permits.
 
While automatic xxencoding of binary files avoids
the {\it email{}} problems noted earlier, for files that
only happen to have tab characters and form feeds,
encoding is unnecessary on Internet-to-Internet
connections.  On the other hand, perfectly normal
\TeX{} files sent in {\it email{}} that goes through
brace-corrupting gateways will be damaged, and may
not be encoded by \TUGLIB{}.  These situations
suggest that the remote user should be able to
control whether encoding is applied, and if so,
what form of encoding.
 
When \FILE{uuencode} can be used safely, it would
probably be more convenient than \FILE{xxencode}
for \unix{} users, because \FILE{uuencode} is already
available on \unix{}.  Other encoding schemes are
available as well, including \FILE{atob} and
\FILE{btoa}, and \FILE{b\-encode} and
\FILE{b\-decode}.  This suggests an addition of new
command verbs to augment the \FILE{send} command.
 
Directory listings are currently only available
through the prior creation of the
\FILE{00dir.\-lst} and \FILE{00tdir.\-lst} files
in each directory.  Perhaps it would be advisable
to generate these dynamically in response to a
\TUGLIB{} command; they would then be guaranteed
to be up-to-date, and disk space would not be used
to store them.  However, those files are also
useful for anonymous \FTP{} retrievals, because it
is not always possible to get more than a bare
list of file names from an \FTP{} \CMD{dir}
command; very often, the file sizes and time
stamps are of interest too.
 
It might also be helpful to have a command like
\CMD{sizeof tex/latex} to return the disk space
requirements of a directory.
 
\section{Acknowledgements}
This work was carried out by the author with the
support of facilities at the Department of
Mathematics at the University of Utah.
 
My deepest thanks go to Eric Grosse and Jack
Dongarra for having written the \NETLIB{} system
and published a paper about it.  I also want to thank Eric
Grosse for making the \NETLIB{} software available
to TUG for modifications to create \TUGLIB{}, and
for keeping a record of everyone who has received
\NETLIB{} so they can get bug fixes.
 
Thanks also go to the TUG Board of Directors for
helping in the testing of \TUGLIB{}, and to Don
Hosek and Joachim Lammarsch for many useful
conversations and electronic exchanges.
 
Finally, we owe an enormous debt to the people who
support and develop the Internet and make it
freely available to a worldwide community
connecting hundreds of thousands of machines, and
perhaps over a million users.  Without their
foresight, many collaborative efforts the world
over would be effectively impossible.



\section{Reference}
{\parindent0pt\frenchspacing
Jack Dongarra and Eric Grosse, 1987, Distribution of Mathematical 
Software via Electronic Mail,
CACM,
30(5),
p.403--407.
}   

\rightline{\sl Nelson~H. F.~Beebe}
\endinput
\ifundefined{SLiTeX}
        \newcommand{\SLiTeX}{{\rm S\kern-.06em{\sc l\kern-.035emi}\kern-.06em
                T\kern -.1667em\lower.7ex\hbox{E}\kern-.125emX}}
\fi 
\bibitem{Dongarra:netlib}
Jack Dongarra and Eric Grosse.
\newblock Distribution of mathematical software via electronic mail.
\newblock {\em Communications of the Association for Computing Machinery},
  30(5):403--407, May 1987.
 
\end{thebibliography}
@//E*O*F tuglib.bbl//
chmod u=rw,g=rw,o=r tuglib.bbl
 
echo x - tuglib.bib
sed 's/^@//' > "tuglib.bib" <<'@//E*O*F tuglib.bib//'
@@string{CACM = "Communications of the Association for Computing Machinery"}
 
@@Article{Dongarra:netlib,
  author =      "Jack Dongarra and Eric Grosse",
  title =       "Distribution of Mathematical Software via Electronic Mail",
  journal =     CACM,
  year =        "1987",
  volume =      "30",
  number =      "5",
  pages =       "403--407",
  month =       may,
}
 
@//E*O*F tuglib.bib//
chmod u=rw,g=rw,o=r tuglib.bib
 
echo x - tuglib.ltx
sed 's/^@//' > "tuglib.ltx" <<'@//E*O*F tuglib.ltx//'
% -*-latex-*-
 
%%  @latexfile{
%%      author          = "Nelson H. F. Beebe",
%%      version         = "1.00",
%%      date            = "30 Aug 1990",
%%      filename        = "tuglib.ltx",
%%      address         = "Center for Scientific Computing
%%                         and Department of Mathematics
%%                         South Physics Building
%%                         University of Utah
%%                         Salt Lake City, UT 84112
%%                         USA
%%                         Tel: (801) 581-5254",
%%      checksum        = "972    4718   34240",
%%      email           = "beebe@science.utah.edu (Internet)",
%%      codetable       = "ISO/ASCII",
%%      supported       = "yes",
%%      docstring       = "
%%                         The checksum field above contains the
%%                         standard \unix{} wc (word count) utility
%%                         output of lines, words, and characters;
%%                         eventually, a better checksum scheme should
%%                         be developed."
%%      }
 
          {$\bullet$}                     
%%  @latexfile{
%%      author          = "Nelson H. F. Beebe",
%%      version         = "1.00",
%%      date            = "28 Aug 1990",
%%      filename        = "tuglib.sty",
%%      address         = "Center for Scientific Computing
%%                         and Department of Mathematics
%%                         South Physics Building
%%                         University of Utah
%%                         Salt Lake City, UT 84112
%%                         USA
%%                         Tel: (801) 581-5254",
%%      checksum        = "56     286    2397",
%%      email           = "beebe@science.utah.edu (Internet)",
%%      codetable       = "ISO/ASCII",
%%      supported       = "yes",
%%      docstring       = "Style file to support tuglib.ltx.
%%                         The checksum field above contains the
%%                         standard UNIX wc (word count) utility
%%                         output of lines, words, and characters;
%%                         eventually, a better checksum scheme should
%%                         be developed."
%%      }
 
\hfuzz=3pt                      %ignore slightly overfull boxes
 
\hbadness=2000                  %ignore most underfull boxes
 
% The standard LaTeX definition of \sloppy sets \tolerance to 10000.
% This gives spacing that is too stretchable, with the result that
% wide white spaces appear in order to give right- justified output.
% We reset the tolerance to a more reasonable value.  We also allow a
% little \emergencystretch (a TeX 3.0 feature) to allow individual
% lines to be slightly longer; if that name is unknown, the code
% avoids its use, so as to work with earlier versions of TeX.
 
% TeXbook, p. 308:
\def\ifundefined#1{\expandafter\ifx\csname#1\endcsname\relax}
 
% Original version from latex.tex:
% \def\sloppy{\tolerance 10000 \hfuzz .5\p@ \vfuzz .5\p@}
\renewcommand{\sloppy}{
    % \emergencystretch is a new feature of TeX 3.0; use it only if known
    \ifundefined{emergencystretch}
      \relax
    \else
      \emergencystretch=3pt
    \fi
    \tolerance=4000             % graph.ltx needs no more looseness than this
    \hbadness=10000             % no underfull messages -- we expect them
}
 
\overfullrule=3pt               % to catch problems in the output.
@//E*O*F tuglib.sty//
chmod u=rw,g=rw,o=r tuglib.sty
 
exit 0