ref: 87288afa5ac476efb3ef1ed40df658427339f13e
dir: /guide/
be sure that each chapter starts at an odd page number.
From: pheras@gmail.com
Date: Mon, 18 Dec 2006 23:08:40 +0100
To: nemo@lsub.org
Subject: libro
p. 113: s/and in well shape/and in good shape/
From: pheras@gmail.com
Date: Tue, 19 Dec 2006 00:03:46 +0100
To: nemo@lsub.org
Subject: libro
en p 114
creo que deberías decir 1 o 2 líneas en esta página sobre note/process
groups: quién pertenece a un grupo... pones un ejemplo con la shell de una
ventana, donde el alumno intuye que todos los procesos creados por la shell
pertenecen al mismo grupo, y por tanto puede extrapolar que un proceso que
crea otros está en el mismo grupo que éstos. Pero estaría bien decirlo. O si
no poner referencia hacia adelante a 7.1 y 7.2.
Por cierto, en p.114 hablas de process group, pero luego en 7.1 y
7.2utilizas la notación plan9: note group. Habría que unificar. Yo
pondría en
114 note group en lugar de process group, y, quizá, no estoy seguro, una
nota a pie diciendo que esto es parecido a process groups en unix.
From: pheras@gmail.com
Date: Tue, 19 Dec 2006 00:08:40 +0100
To: nemo@lsub.org
Subject: libro
p116
However, Thisfunction
From: pheras@gmail.com
Date: Tue, 19 Dec 2006 00:17:18 +0100
To: nemo@lsub.org
Subject: libro
p117
Cambiar
It requirestheprocesstoinstall ahandler for theinterruptednote
por ... the interrupt note
No queda claro que el manejador obtiene "interrupt" en *msg, mientras que
en tras una nota tratada en rerrstr se obtiene el string "interrupted"
En el código es lo que parece, pero no en el texto
From: pheras@gmail.com
Date: Tue, 19 Dec 2006 00:51:43 +0100
To: nemo@lsub.org
Subject: libro
p124
s/doesoccurs/does occur/
Sobre la semántica de close.
En p. 100, dices:
Atthispoint, thedescriptorwhosenumberisinfd
isnolongernecessary, andcanbeclosed.
Close no cierra descriptores, sino ficheros. Creo que aquí hay que rehacer
algo, pues puede liarse el lector.
Convendría explicar que el close después de dup cierra el fichero, en lo
que a ese descriptor se refiere, pero el mismo proceso mantiene abierto ese
mismo fichero a través del otro descriptor.
O quizá bastaría con aclararlo en el capítulo 3, mostrando un programa que
abre 2 veces para lectura/esc un mismo fichero, y luego cierra el fichero
una vez, pero sigue pudiendo leer a través del otro descriptor.
Aunque creo que habría que reincidir en p.100 al volver al dup, pues tras el
dup son 2 descriptores apuntando al mismo chan, y tras cerrar el fichero a
través de un descriptor, sigue estando abierto a través del otro descriptor.
en p. 103 cambiar
whichisequivalenttoadup(1,2)
por
.... dup(2,1)
programa pipeto p109
Está muy bien hacerlo genérico, pero puede complicar la comprensión del uso
de pipe el poner rc como proceso que ejecuta grep, en lugar de que se
ejecute directamente.
De hecho en la figura 5.3 no aparece rc...
p104: andwhattocheckthatout,
Thisfileinisanother invention
p102: Iftheprogramhadwrite
x
If the program had written
p. 97
s/did now write anything/did not write anything/
El problema 6 y el 7 de p.95 no se entienden bien.
La respuesta al 6 es:
#!/bin/rc
/bin/rc $*
Pero qué siginifca tu pregunta: "Can you specify..."
Y por tanto el problema 7, pues que tampoco lo entiendo.
problema 5 p95:
¿quieres que devuelva el msg del struct Waitmsg devuelto por wait, ¿no?
Despista el Hint. La respuesta al Hint es "shell", pero, ¿cómo ayuda saber
responder el hint a resolver el problema?
el problema 4 de la p95 no se puede resolver sin hacer man wait
En la p91 no explicas que el array time te da esos 3 tiempos que pides.
Aunque se intuye :-)
p.30, programa take.c
¿no deberías poner la primera vez que aparece "nil" qué es?
p93, l-3: s/loader/loader\/linker/
> > p87: cambiar "dangerous" por "exciting and dangerous"
en p75 dice:
Asyoucansee, bothBtermandBflushreturnaninteger.
Pero es la pprimera vez que se habla de Bflush. Habría que decir qué hace
justo ahí.
en programa lsdot.c, p. 67 falta close
p.63, rm.c: hay que poner argc, argv
> cambiar en el código 0664 por 0644
andrecordthewhat thefilewasopen
Andthefilewereto
where, no were
A couple of small things from chapter 12:
Section 12.2 - I believe Rune is defined in u.h rather than libc.h
Section 12.5, the text of black.c doesn't quite match the surrounding
discussion - the sleep call is for 5 seconds rather than the 10 seconds
in the text, and the call to draw is:
draw(screen, rect, black, nil, Pt(rect.min.x+20, rect.min.x+20);
rather than
draw(screen, screen->r, display->black, nil, ZP);
as discussed in the text on p298.
: Chapter 8 - in general, my shell doesn't print the double prompt on
: continuation lines, just indents them - can this be configured somewhere?
Most times you use the word "miss" I think you probably mean "lack",
maybe sometimes "omit" (I suspect you are translating from "falta"?).
As a Windows & Mac user, I found the user-interface style of rio & acme
initially quite alien, and I think more description of this could be
helpful, depending on the background of your students – eg the use of
the three button mouse in rio, and particularly how the "Send" command
allows you to treat the whole window as a kind of scratchpad to assemble
commands, and does away with the need for the normal history mechanisms
using command line recall and editing. I was very slow to appreciate
this aspect of the system. I think it would also be good to have more
info on using acme – maybe by including the acme paper as an appendix.
The way that scrolling and the cursor keys work is initially baffling
and frustrating to a Windows user, and I’ve still not mastered chording.
I initially had trouble in following the discussion in Section 5.5,
because I thought that you were saying that the mail program internally
used the technique in pipeto.c, and I couldn’t see why it should. I
think I was misled because of the way the sentence "This can be used for
many things" was immediately followed by "For example, the mail program
...", which led me to think that the mail program was an example of a C
program of the type mentioned in the first sentence, rather than a
command being executed by such a program.
In the listing rforkls.c in Section 1, the first flag to rfork should
presumably be RFFDG, not RFCFDG?
In section 7.3, I was not clear of the exact meaning of "After executing
copy, the environment is not yet known to our shell. The reason is that
the shell caches environment variables." Will the new environment ever
become known to the current instance of the shell, or is it the case
that environment variables are loaded only once, when the instance of
the shell is started, so that the current instance will never see the
new variable? If the latter, then "not yet" is a bit of an understatement.
In Section 7.5, you introduce the mount command without describing the
–c option, but then use this option the first time you use the command –
I would have liked to see it explained here, but then maybe you’re
trying to get your students to use the manual, rather than spoon-feeding
them all the time.
When you first introduced "bind" in the context of:
bind –c /n/whale /n/other
my immediate response was, "why on earth would I want to do that?" You
make the uses of bind very clear later on in section 7.10 – perhaps a
comment and a forward reference might be helpful?
On p150, the "grep whale /proc/843/ns" and "ns" commands produce output
containing the device name "#s", rather than the directory name "/srv"
which you used originally, and you don’t introduce devices until section
7.7 – a comment and a forward reference could be helpful. Also, more
detail on the differences between the output from the two commands would
be useful – is it just the single quotes around #s/tcp!whale!9fs? It's
surely a matter of taste whether they make the output prettier.
In section 7.7, I think I understood the discussion, but found myself
wondering how exactly it got us any further – we started by wondering
how anything got mounted at /, and I ended up wondering how #/ managed
to provide any files – this could well just be stupidity on my part,
combined with a complete ignorance as to how device drivers work.
The sandbox program in section 7.11 has a few problems. It should test
for argc != 3, and in this context the final argument to fprint should
be argv[0] rather than argv0, which has not been initialised. The
arguments to execl should be argv[2], not argv[1]. It may be worth
pointing out that if you are foolish enough not to bind a directory
containing rc in "sandbox" – as I was – then the example will fail to
execute /bin/rc.
sam, ed
yesterday
process priorities
upasfs
timezone, ctim
pipefile
- que llevas usando Plan 9 para docencia más de 5 años, con un buen
resultado.
- En el item 5 de "outstanding features", hacer hincapié en que los
sistemas populares (pe Linux) cada vez incluyen más conceptos de Plan 9,
como /proc, varios ns, bla bla. Por lo tanto, el libro no trata sobre un
bicho raro cuyos mecanismos no se aplican en sistemas populares, justo
lo contrario: cada vez tiene más influencia.
- que el otro libro que has escrito ha tenido éxito entre los alumnos
y la comunidad de Plan 9. Dale bombo. Pon la url.
- Austing --> Austin
Solo una cosa. Creo que tanto en audiencia como en promocion no has
vendido suficientemente bien, piensa que esa gente quiere vender libros.
Por ejemplo, en promotion, pondria algo como:
There is a growing and widespread interest in Linux and Unix like systems
and operating systems in general.
More and more people are interested in how a operating system works. Unix
systems have grown to be too complex to be understood by only one person.
The intended audience would be anyone willing to learn how a full
fledge operating system works which includes hobbyists, professional
programmers and students
and CS an CE students. This is specially true for people interested in
Unix since the system it is based on one which was written by the
people who designed and implemented it after realizing their mistakes.
Algo parecido en promotion. Parece que solo se podria promocionar en
9fans y puede que en otras comunidades de linux. Creo que deberias ser
mas agresivo. El libro no es solo plan 9, es más general.
From: pheras@gmail.com
Date: Tue, 19 Dec 2006 20:06:44 +0100
To: nemo@lsub.org
Subject: libro
problema 2 p129
1. en open falta cerrar comillas
2. Pregunta: la respuesta a lo que preguntas es "porque 0 no debe tener
permiso de escritura"?
Me ha costado poco ver dónde está el código: boot.c
Explícame este problema
From: pheras@gmail.com
Date: Tue, 19 Dec 2006 20:23:18 +0100
To: nemo@lsub.org
Subject: libro
En problema 3 p 129:
Respuesta a 1ª pregunta: "porque puede llegar una nota durante la ejecución
de open, y si hay un manejador para la misma, después se llama a dup,
quedando apuntando 0 a donde fd: al aire"
¿sí?
En el texto de la siguiente pregunta de este problema s/notice/NOTICE/
Respuesta: </NOTICE
From: pheras@gmail.com
Date: Tue, 19 Dec 2006 20:35:58 +0100
To: nemo@lsub.org
Subject: libro
problema 4 p 129
Quieres que hagan un pipe, llamen a read sobre fd[0] y vean que está en
PRead?
From: pheras@gmail.com
Date: Wed, 20 Dec 2006 00:10:34 +0100
To: nemo@lsub.org
Subject: libro
p. 131: s/theremanyprograms/there are many programs/
s/sometypeofwireortheairforradiocommunication/copper wires, optical fibers
or the air for radio communication/
From: pheras@gmail.com
Date: Wed, 20 Dec 2006 00:17:18 +0100
To: nemo@lsub.org
Subject: libro
p. 131:
s/anethernet network(just sometypeof
cablingandconventions)./an ethernet network: the most used link technology
(hw+protocols) for digitally send bits over wires.
From: pheras@gmail.com
Date: Wed, 20 Dec 2006 00:30:20 +0100
To: nemo@lsub.org
Subject: libro
p 131
s/Machinesattachedtothewirehaveaddresses/In some link technologies like
ethernet, machines attached to the wire have link addresses/
s/identifydifferentmachinesattachedtothewire./identify different machines
attached to the same wire/
From: pheras@gmail.com
Date: Wed, 20 Dec 2006 00:31:51 +0100
To: nemo@lsub.org
Subject: libro
p.131
s/wireyour machineisattachedat/wire your machine is attached at, and even if
its particular link technology does not provide link addresses!/
From: pheras@gmail.com
Date: Wed, 20 Dec 2006 00:35:21 +0100
To: nemo@lsub.org
Subject: libro
p.131
s/it knowshowtoreachanymachine, givenitsaddress./it knows how to make
information reach its destination given its network or IP address/
From: pheras@gmail.com
Date: Wed, 20 Dec 2006 00:36:45 +0100
To: nemo@lsub.org
Subject: libro
p. 131
s/TheinterfacefortheIPnet-
workinPlan9issimilartotheonewesawforEthernet/The interface for accessing the
IP protocol in Plan 9 is similar to the one .../
From: pheras@gmail.com
Date: Wed, 20 Dec 2006 00:38:48 +0100
To: nemo@lsub.org
Subject: libro
p.132
s/provideanabstractionsimilar/provide a service similar/
From: pheras@gmail.com
Date: Wed, 20 Dec 2006 00:40:26 +0100
To: nemo@lsub.org
Subject: libro
p.132
TheIPdevicedriver
The TCP/IP software
Pág. 48, se menciona que rabbits.c es peor que diehard.c y que
crea procesos de forma exponencial, pero no es así, tal y como
está rabbits.c, todos los hijos mueren inmediatamente, por lo que
la creación de procesos es lineal y además es sencillo de
terminar, basta con matar al padre original.
rabbits.c es:
while(fork())
;
exits(nil);
Debería ser algo como:
while(fork() >= 0)
;
exits(nil);