Warning: this is an htmlized version!
The original is across this link,
and the conversion rules are here.
\input texinfo

@c This is the `doc/eev.texi' file of GNU eev.
@c Copyright (C) 2004 Free Software Foundation, Inc.
@c Copying and distribution of this file, with or without modification, are
@c permitted provided the copyright notice and this notice are preserved.
@c Author and version: Eduardo Ochs, 2005feb16

@c Note that this document is still very preliminary.

@c (setq ee-hyperlink-prefix "@c ")
@c (setq ee-comment-prefix "@c")
@c (find-texinode "Emacs Editing")
@c (find-texinode "Updating Commands")
@c (find-texinode "Updating Requirements")
@c (find-eev "README")

@c (find-texinode "Concept Index")
@c (find-node "(texinfo)makeinfo options" "\n`--html'\n")

@c (find-node "(texi2html)")
@c (find-node "(texi2html)whytexi2html")

@ignore

# (find-status   "texi2html")
# (find-vldifile "texi2html.list")
# (find-udfile   "texi2html/")

(defun ei ()
  (interactive)
  (find-estring
   (format "%s\n%S"
           (find-sh0 "cd ~/eev-current/doc/; makeinfo eev")
           ' (find-node "(eev)Top")
           )))

cd ~/info/
ln -sf ~/eev-current/doc/eev.info .

@end ignore

@c (find-sh "cd ~/eev-current/doc/; makeinfo eev && cp -v eev.info ~/info/")
@c (find-sh "cd ~/eev-current/doc/; makeinfo eev")
@c (find-sh "rm -Rv ~/eev-current/doc/eev/; cd ~/eev-current/doc/; makeinfo --html eev")
@c (find-eevnode "")

@c (eev     "cd ~/eev-current/doc/; makeinfo eev")
@c (eev     "cd ~/eev-current/doc/; texi2dvi eev.texi && rexdvi eev.dvi")
@c (find-fline "~/eev-current/doc/eev.info")
@c (find-node "(eev)Top")

@c (remove-from-keymap texinfo-mode-map ?\")
@c (define-key texinfo-mode-map "\"" 'texinfo-insert-quote)
@c (find-efile "textmodes/texinfo.el" "if texinfo-mode-map" "texinfo-insert-quote")


@setfilename eev.info
@settitle eev

@node    Top, Introduction, (dir), (dir)

To edit this: (find-eevfile "doc/eev.texi")

@menu
* Introduction::                What are eev and e-scripts.
* Reading E-scripts::           
* Installing eev::              Installing, uninstalling, and how to run eev
                                without installing it.
* Executing Regions::           
* Hyperlinks::                  Elisp one-liners used as hyperlinks.
* Technicalities::              
* Projects::                    
* Two Halves::                  
* Concept Index::               This index has two entries.

@detailmenu
 --- The Detailed Node Listing ---

Introduction

* History Editors::             Many programs include tiny editors.
* E-scripts::                   What are eev and e-scripts.
* Between Code and Actions::    Everything can be automated, in theory.
* Sharing E-scripts::           Reasons for sharing information.
* The Eev Manifesto::           The original Eev Manifesto, written in 1999.
* Basic Examples::              
* No Autodetection::            
* Bounded Regions::             

Installing eev

* Changes to rcfiles::          
* Automatic Installation::      
* Automatic Uninstallation::    
* make preinstall-rc::          Preparing the new rcfiles.
* make diff-rc::                Comparing the new rcfiles with the current ones.
* make install-rc::             Installing the new rcfiles.
* Running Without Installing::  Trying eev without installing it.

Executing Regions

* Kinds of Regions::            Emacs's region, bounded regions, lines, sexps.
* Kinds of Actions::            Saving, evaluating, executing, signaling.
* eesteps::                     Loading an executing series of Elisp actions.
* Defaults for Bounded Regions::  Delimiters, action, how to highlight.
* Flashing::                    Highlighting the region temporarily
* Channels::                    Sending lines through communication channels.

Hyperlinks

* Kinds of Hyperlinks::         To files, info, man, Emacs structures, etc.
* Pos-Spec Lists::              Pointing to a position in a page.
* code-c-d::                    Generating many hyperlink functions at once.
* Generating Hyperlinks::       Generating many hyperlinks at once.

Technicalities

* Glyphs::                      
* Compose Pairs::               
* ee-expand::                   

Projects

* Tutorials for Languages::     
* Internet Skills for Disconnected People::  
* Quick References::            

Splitting Eev in Two Halves (and Their Frontier)

* Saving::                      Saving temporary scripts.
* Links::                       
* Actions to Code in Emacs::    
* Hybrid Functions::            
* More on Hyperlinks::          

@end detailmenu
@end menu

@node    Introduction, Reading E-scripts, Top, Top
@chapter Introduction

@menu
* History Editors::             Many programs include tiny editors.
* E-scripts::                   What are eev and e-scripts.
* Between Code and Actions::    Everything can be automated, in theory.
* Sharing E-scripts::           Reasons for sharing information.
* The Eev Manifesto::           The original Eev Manifesto, written in 1999.
* Basic Examples::              
* No Autodetection::            
* Bounded Regions::             
@end menu

@node    History Editors, E-scripts, Introduction, Introduction
@section History Editors

Many interactive programs - for example shells and interpreters for
certain programming languages - keep a history of the last commands
that they received.  This allows the user to edit these commands and
execute them again, with or without changes. Let's call these editing
facilities ``history editors''.

The main idea of eev is that if we routinely use programs A, B, C and
D then we don't need to restrict ourselves to A, B, C and D's built-in
history editors: Emacs can function as an ``auxiliary history editor''
for them.  We can use a single running Emacs to edit and send commands
for all these programs.  And when we use Emacs 



Using a single running Emacs can edit commands for

 - a global one, in fact [capable of dealing with several
programs, unifying the histories, supporting comments and hyperlinks]

These history editors have several deficiencies. [They are different
from program to program; they don't have all the facilities of a
full-blown text editor]

[Mention comments and hyperlinks; stuff for several programs in the
same file]

[Saving history; editing history; keeping a log; several separate
logs]

@c Vários programas interativos - por exemplo shells e certas
@c linguagens de programação - mantêm um histórico dos ultimos
@c comandos recebidos, e permitem que o usuário edite estes comandos e
@c os execute de novo, com ou sem alterações; às vezes estes programas
@c até contêm pequenos editores de texto embutidos.

@c A idéia do eev é não nos restringirmos somente a estas ferramentas
@c simplificadas de edição de comandos - que ainda por cima são
@c diferentes de programa a programa. Com o eev o Emacs passa a poder
@c ser utilizado como um "editor universal de comandos" - os e-scripts
@c são como se fossem históricos globais de comandos dados para o
@c sistema, mas podendo ser editados como arquivos de texto comuns,
@c podendo conter todo tipo de comentário e de hiperlink, e podendo
@c conter comandos para vários programas e linguagens.



@node    E-scripts, Between Code and Actions, History Editors, Introduction
@section E-scripts

Eev is a library that adds support for "e-scripts" to Emacs. E-scripts
are files of "executable notes" about how to do certain computer tasks.

A typical e-script is a plain text file in which some parts are
written in human languages, like English, and other parts are blocks
of code in certain programming languages or sequences of commands for
certain computer programs.  A single e-script file may contain blocks
of code and commands for several different languages and programs.

We are used to the idea that if a file `myescript.e' contains
executable code then there is a way to execute it; maybe there's a
program, `myprog', such that running `prog myescript.e' will execute
the e-script.  But the situation with e-scripts is quite different
from that.

There's no way to execute a whole e-script file at once. Instead, the
user selects interactively a region of the e-script while editing it
in Emacs, and then ask Emacs to execute it in some way. It is always
up to the user to decide in which way to execute the selected region -
or, in other words, to determine ``in which language the contents of
the region are written''. Emacs never tries to guess the language - or
the region - by itself.

Executing means different things for different languages and for
different programs.  If the region is to be executed as Emacs Lisp
code, then executing it means applying two Elisp functions to the
contents of the region: `read' and `eval'.  If the region is to be
sent as a series of commands to an external shell, then it is saved as
a temporary file - a ``temporary script'' -, so that then we can
switch to the shell prompt and execute that temporary script. And if
the region is to be sent as input to some external interactive program
with which Emacs has established a certain kind of ``communication
channel'', then executing that means saving the region to a temporary
file and then sending a signal to the external program to tell it that
more input has just been made available.

There's one kind of executable block that is especially common:
one-liners in Elisp, being used as hyperlinks. For example: in

@example
# (find-node "(make)Automatic" "$@@")
# (find-man "1 bash" "echo [-neE]" "\\t")
cd /tmp
touch first; sleep 1; touch second
@{ echo 'first: second'
  echo -e '\t@@echo $@@ is older than $<'
@} > Makefile
make
@end example

the whole thing - eight lines - can be executed as shell code, and
then the first two lines will be treated as comments; but we can also
put the cursor at the end of any of the `#' lines and type a key
sequence - `M-e' - that means ``go to the end of the line and then
execute the Elisp expression that comes before the cursor''; then the
`#' at the beginning of the line will not be considered, only the
thing between parentheses - the `#' will be beyond the limits of the
region - and the effect will be that Emacs will open a certain manpage
or Info page and jump to a specific position.  That is, executing that
expression works like following a hyperlink.

[mention defaults, and few keystrokes].



@c ...

@c Old stuff:

@c An e-script is a file containing notes about how to do a certain
@c computer task. Some parts of an e-script are written in English or
@c in other human languages, and are not intended to be interpreted by
@c the computer in any way; other parts contain logs of the
@c interactions with the programs that were involved in the task, in a
@c form that allows not only inspection and textual modification but
@c also controlled ``re-execution'' - that is, repeating that
@c interaction in another moment, with or without modification, from
@c the steps registered in the log, without the need to repeat each
@c keystroke or mouse-click by hand.

@c So: e-scripts are a way of taking ``executable notes'' of what you
@c do. If you interact with the system using Emacs and Eev then notes
@c in e-script form are produced more or less automatically, and
@c redoing things that you've already done before becomes trivial.


@node    Between Code and Actions, Sharing E-scripts, E-scripts, Introduction
@section Between Code and Actions

An ``action'' can be ``recorded'' as text by writing down code that
produces that action when executed - remember that code is text! The
transformation

@example
               ``action": effect
    code   ->     of executing
                   the code
@end example

is natural for the computer; the opposite direction,

@example
                  code that
   action   ->    produces
                 that action
@end example

is not. For many actions we may know how to do that by hand, and for
some kinds of actions there are tools to automatically generate the
corresponding code; we will discuss some of these tools later (section
"Following and Creating Elisp Hyperlinks", below).

@example
# (find-eevfile "README" "If our \"action\" is")
@end example

[Ways to trigger a stored action]


@node    Sharing E-scripts, The Eev Manifesto, Between Code and Actions, Introduction
@section Sharing E-scripts

[I should write something new here, but for the moment the original
``Eev Manifesto'' should do, with some added comments]

[I believe that in a few decades...]

[You are not forced to share your e-scripts.]

[If you are releasing a program...]



@node    The Eev Manifesto, Basic Examples, Sharing E-scripts, Introduction
@section The Eev Manifesto



The eev Manifesto
=================

I)

Do you remember your first Unix days, when you were always asking
yourself questions like "how do I compile this?", "how do I install
this stuff?", "and now, how do I run it?", "how do I generate html
from this file?"?

Do you remember how you marvelled at the first time you ran a "make"
on a package and you saw echoed each of the commands that were used to
generate binaries and docs from the source? Have you had the same
feeling I've had, that by understanding those commands we would have
access to the tricks that the owner of the package knew, and that we'd
learn which were the "important" programs and in what way the experts
used them?

II)

Everybody is fluent in only a small fraction of all Unix commands. If
you could "listen" to how the Unix gurus "speak" to their machines you
would learn which "words" are related to solving a particular task,
and learn how they fit in "sentences". By checking the "dictionary
entries" for them (i.e., manpages, info pages, READMEs, source code,
etc) you could learn the real meaning of them. But then you'd be
learning Unix by immersion, from real use, instead of having to rely
only on "textbooks", "dictionaries" and sometimes "Rosetta stones",
"graffitis on toilet walls" and "old newspapers".

The fact is that you can make a record of how you "speak" Unix, and
more, you can become a lot more productive if you do so. Many tasks
consist on short fixed sequences of commands: connecting to your ISP
via modem, unpacking a source package and recompiling it, printing a
text file in two-column mode, and so on. The trick is that with some
functions defined in eev.el you can write these sequences of commands
in a plain text file, then mark a block of this file with your editor
(which must be Emacs for this to work), then tell a shell to execute
only the commands in that block; in this way you can easily execute
only portions of what would otherwise have to be a monolythic script;
this is great for when you're not sure if everything works, or if you
just want to do some steps. Also, it would be easy to change bits of
the "script" before execution, as you'll be doing things from inside
an editor.

There is also another trick. Scripts allow "comments", that is, lines
that are not executed; generally the convention is that lines that
start with "#" are comments. So we can place lines like these in a
script:

  # (find-enode "Picture" "C-c c-c")

The expression "(find-enode "Picture")" is an Emacs command (in Lisp)
that means: open the "info page" called "Picture", from the Emacs
manual, treating it as a file; then place the cursor after the first
occurrence of the string "C-c c-c" there. To execute this command,
i.e., to make emacs go to that point on that page, you just have to
put the cursor after the ")" and type C-x C-e. There are many other
similar commands, that will open specific files or specific manpages,
with various ways of searching. These "hyperlinks" are especially
interesting if you want to record from where you got some piece of
information relevant to the "script" you're writing; this is very
useful, for example, on "scripts" you haven't finished and on which
you plan to work more someday, or on scripts you want to send to
someone.

Just as hyperlinks have many flavors, the function that executes a
block also has variants: a block can be executed by the Tcl
interpreter instead of the shell, or can be run through LaTeX, just to
pick some examples. See the docs on eev.el
(<http:/angg.twu.net/eev-current/README.html>).

III)

I have placed essentially all my "scripts" written in this way (I call
them "e-scripts") in a public place (<http:/angg.twu.net/e/>). They
contain almost everything I know about Unix.

If you like this idea, please get in touch, send comments, ask
questions -- about e-scripts or questions whose answer could become an
e-script chunk -- or send me your e-scripts when you have some, or
even ask for help on setting up your own e-script collection or
e-script public site... anything! By asking good questions you can
help me make the documentation get better.

I really want to make this e-scripts idea spread. Learning Unix -- or
simply more Unix -- could be made easier for everybody... please help!
E-scripts are funnier to use, and easier to write, than texts that
tell everything in terms of "press this, do that". A lot of effort and
money are being invested now on these kinds of text, and they're often
very depressing. Let's try to save the world from them, at least a
bit, and maybe this money will be directed to better things. And
teaching people Unix tricks will be both easier and funnier.

Eduardo Ochs, 1999dec18.




@node    Basic Examples, No Autodetection, The Eev Manifesto, Introduction
@section Basic Examples: Shell Commands and Hyperlinks

If our ``action'' is to enter these commands in a shell,

@example
cd /home/backups/
tar -tvzf persephone-2002jun10.tar.gz > ~/.files
grep books.db ~/.files
@end example

@noindent then we can record that by just pasting the commands to Emacs (by the
way, in this case it is the way of playing back the commands that is
not obvious! See the screenshot below for the trick). But now suppose
that we have just found a piece of interesting information in the
documentation (in Info) for Zsh, in the node called ``Process
Substitution'', around the first occurrence of the string
@code{=(...)} in the text - how do we create a ``hyperlink'' to that,
that is, one whose action when ``followed'' will be to take us to that
page and to put the cursor at that point? The elisp expression below
does the job:

@example
(find-node "(zsh)Process Substitution" "=(...)")
@end example

We can execute it by placing the cursor after it and typing @key{C-x
C-e}, or by placing the cursor anywhere in the line and typing
@key{M-e}. Note: @key{C-x C-e} is standard in Emacs, while (this
behaviour for) @key{M-e} is defined in eev; @code{find-node} is from
eev, but Emacs comes with several functions that make sense as
hyperlinks, for example:

@example
(find-file "/etc/fstab")
(info "(zsh)Process Substitution")
(info "(emacs)Lisp Eval")
(man "ls")
(describe-key "\C-x\C-e")
(describe-function 'eval-last-sexp)
(where-is 'eval-last-sexp)
@end example

These ``hyperlink'' functions that are native to Emacs do not accept the
extra arguments specifying where to move the cursor to, but all the
functions in eev do (these extra arguments are called ``pos-spec-list"s
in eev, by the way).

Note that these ``elisp hyperlinks'' can appear inside comments in code
for other languages:

@example
# (find-node "(zsh)Process Substitution" "=(...)")
function catcat () @{ echo "Temp file: $1"; cat $1; tac $1 @}
catcat =(echo foo; echo bar; echo plic)
@end example

This is a valid block of shell commands - code for one language can
appear inside code for another language.


@node    No Autodetection, Bounded Regions, Basic Examples, Introduction
@section Languages are never auto-detected

Eev does not try to recognize automatically what are the limits of a
block of code, or in which language a block was written. This is not a
bug or a missing feature --- it is a deliberate design decision.
(Actually in the beginning there was no autodetection because I didn't
know how to implement it; but then I found that having no
autodetection had certaing advantages

It would be very hard to make a system supporting several languages if
the limits of each block had to be found automatically, and if each
language had to identified.

In eev the boundaries of the block and the language are always
selected @i{interactively}: the user selects the block and then issues
a command meaning ``execute the block as code in the language
such-and-such'' (or, in the worst case, ``execute the block as code in
the default language'').

Thanks to that the implementation of eev is very simple -
and adding support for additional languages is quite easy.

[To do, maybe: include a list of currently supported languages, and
write a (big) section about Interacting with Unsupported Things].




@node    Bounded Regions,  , No Autodetection, Introduction
@section Bounded Regions


The commented screenshot below illustrates one way of sending commands
to a shell - by putting the cursor inside a ``delimited region'' and
typing @key{F3}, then going to shell and running `ee' there.

@example
(find-eimage0 "doc/shot-f3.png" 1 nil "40%")
@end example

It is also possible to do something similar with regions that are not
bounded by delimiters; just mark the region in the usual way in Emacs,
by setting the ``mark'' and moving the ``point'', then type M-x eev.

One variation: if you mark a region and issue @key{M-x eelatex}
instead of @key{M-x eev} then the region will be ``sent to @LaTeX{}''
instead of to the shell; the region will be written in a temporary
.tex file, and the temporary script will receive commands to run latex
on the temporary .tex file and display the resulting .dvi file; after
@key{M-x eelatex} you switch to a shell and run `ee' there.

@example
(find-enode "Mark")
@end example

By the way, the name `eev' stands for something like
"Emacs-execute-(in a shell)-verbosely". It was one of the first Emacs
Lisp functions that I wrote, and as I was executing it several
hundreds of times a day it had to have a very short name that would be
very easy to type.



@node    Reading E-scripts, Installing eev, Introduction, Top
@chapter Reading E-scripts

Understanding an e-script file; examples

@node    Installing eev, Executing Regions, Reading E-scripts, Top
@chapter Installing eev

Installing eev means not only making Emacs load eev's functions, but
also adding commands to several programs -- Bash, Zsh, GDB, Tclsh,
Python, Perl, Fvwm, etc -- to make it easy to execute the temporary
scripts from them.

[The added command is usually called `ee'. Explain the security
issues.]

@example
(find-eevfile "README")
(find-eevfile "INSTALL")
(find-eevfile "Makefile")
(find-eevrcfile "README")
(find-eevrcfile "tmp/")
(find-eevrcfile "tmp/README")
@end example

@menu
* Changes to rcfiles::          Adding an `ee' command to .bashrc, .gdbinit...
* Automatic Installation::      The easy way.
* Automatic Uninstallation::    The easy way can be undone easily.
* make preinstall-rc::          Preparing the new rcfiles.
* make diff-rc::                Comparing the new rcfiles with the current ones.
* make install-rc::             Installing the new rcfiles.
* Running Without Installing::  Trying eev without installing it.
@end menu

@node    Changes to rcfiles, Automatic Installation, Installing eev, Installing eev
@section Changes to rcfiles

@node    Automatic Installation, Automatic Uninstallation, Changes to rcfiles, Installing eev
@section Automatic Installation

@node    Automatic Uninstallation, make preinstall-rc, Automatic Installation, Installing eev
@section Automatic Uninstallation

@node    make preinstall-rc, make diff-rc, Automatic Uninstallation, Installing eev
@section make preinstall-rc

@node    make diff-rc, make install-rc, make preinstall-rc, Installing eev
@section make diff-rc

@node    make install-rc, Running Without Installing, make diff-rc, Installing eev
@section make install-rc

@node    Running Without Installing,  , make install-rc, Installing eev
@section Running Without Installing



@node    Executing Regions, Hyperlinks, Installing eev, Top
@chapter Executing Regions

@menu
* Kinds of Regions::            Emacs's region, bounded regions, lines, sexps.
* Kinds of Actions::            Saving, evaluating, executing, signaling.
* eesteps::                     Loading an executing series of Elisp actions.
* Defaults for Bounded Regions::  Delimiters, action, how to highlight.
* Flashing::                    Highlighting the region temporarily
* Channels::                    Sending lines through communication channels.
@end menu

@node    Kinds of Regions, Kinds of Actions, Executing Regions, Executing Regions
@section Kinds of Regions

Current kinds of regions: Emacs's region (example, M-x eev), bounded
regions (M-x eev-bounded or <f3>), sexp before point (M-E and <digit>
M-E), sexp before eol (M-e and <digit> M-e), current line (<f9>, i.e.,
eechannel-do-this-line), no region (for eesteps-do-step).

@node    Kinds of Actions, eesteps, Kinds of Regions, Executing Regions
@section Kinds of Actions

Building blocks: executing as Lisp, saving to a file, running through
a shell immediately, flashing, sending a signal, executing one of the
stored steps.

@node    eesteps, Defaults for Bounded Regions, Kinds of Actions, Executing Regions
@section eesteps

Try this: execute:

@example
(eesteps '("C-x b * scratch * RET"
	     "foobar"
           "3*<left>"
           (insert "!")))
@end example

then type @key{F12} four times, slowly.  Emacs will:

@enumerate
@item
switch to the buffer "*scratch*"
@item
insert "foobar"
@item
move the point left three characters
@item
insert "!".
@end enumerate

Typing another @key{F12} will give an error ("No more steps").

Here's how this works, internally: the variable `eesteps-list' holds a
list of "steps" to execute, and the variable `eesteps-pos' holds a
pointer (a number) saying where we are in that list; the function
`eesteps' receives a list of steps, stores that in `eesteps-list' and
resets `eesteps-pos' to 0.

The steps in `eesteps-list' can be either lists -- and in this case
executing the step means evaluating the list -- or strings; strings
are interpreted as sequences of keys, written in the verbose format
used by `edmacro-mode'.

@example
(find-eev "eev-dev.el" "eesteps")
(find-eev "eev-dev.el" "eek")
(find-eev "eev-dev.el" "eek0")
(find-efile "")
(find-efile "edmacro.el")
@end example

The function @code{eek} is a kind of ``hyperlink to executing a key
sequence''; it is used by @code{eesteps-do-step}. @code{eek0} is
similar but uses Emacs's native format for keyboard macros, which is
usually more compact and harder to read.

[Problem: eek and eek0 run their strings as macros, but macros runs
in a kind of non-verbose mode - M-w doesn't give visual feedback]


@node    Defaults for Bounded Regions, Flashing, eesteps, Executing Regions
@section Defaults for Bounded Regions

@node    Flashing, Channels, Defaults for Bounded Regions, Executing Regions
@section Flashing

@node    Channels,  , Flashing, Executing Regions
@section Channels

This is a new feature that is très chic: controlling several programs
at the same time. Click on the screenshot to enlarge.

@example
(find-eimage0 "doc/shot-f9.png" 1 nil "40%")
(find-eev "eegchannel")
@end example


News: we can use the same trick - save the "line" in a file, then send
a SIGUSR1 to the process responsible for the channel - with strings
that are not necessarily lines and with controlling processes that are
no eegchannel; see:

@example
(find-eevex "python.e" "eechannel-python")
(find-eevex "tcltk.e"  "eechannel-tcl")
@end example





@node    Hyperlinks, Technicalities, Executing Regions, Top
@chapter Hyperlinks

@menu
* Kinds of Hyperlinks::         To files, info, man, Emacs structures, etc.
* Pos-Spec Lists::              Pointing to a position in a page.
* code-c-d::                    Generating many hyperlink functions at once.
* Generating Hyperlinks::       Generating many hyperlinks at once.
@end menu

@node    Kinds of Hyperlinks, Pos-Spec Lists, Hyperlinks, Hyperlinks
@section Kinds of Hyperlinks

@node    Pos-Spec Lists, code-c-d, Kinds of Hyperlinks, Hyperlinks
@section Pos-Spec Lists

@node    code-c-d, Generating Hyperlinks, Pos-Spec Lists, Hyperlinks
@section code-c-d

@node    Generating Hyperlinks,  , code-c-d, Hyperlinks
@section Generating Hyperlinks


@node    Technicalities, Projects, Hyperlinks, Top
@chapter Technicalities

@menu
* Glyphs::                      
* Compose Pairs::               
* ee-expand::                   
@end menu

@node    Glyphs, Compose Pairs, Technicalities, Technicalities
@section Glyphs

To type the red stars in these delimiters, use C-q C-o. This inserts
a "literal C-o" in the text, whose ascii code is 15; it would
normally be displayed as "^O", but we change that to a red star
using "glyphs" (see the file glyphs.el). glyphs.el defines several
other non-standard glyphs, like "«" and "»". We can enter those
typing M-, < < and M-, > >, respectively. The function associated to
`M-,', `compose-two-keys', is defined in compose.el. Neither glyphs
nor these "compose pairs" would be strictly needed for eev, but I
find them convenient.

@example
(find-eev "glyphs.el")
(find-eev "compose.el")
@end example


@node    Compose Pairs, ee-expand, Glyphs, Technicalities
@section Compose Pairs

@node    ee-expand,  , Compose Pairs, Technicalities
@section ee-expand


@node    Projects, Two Halves, Technicalities, Top
@chapter Projects

@menu
* Tutorials for Languages::     
* Internet Skills for Disconnected People::  
* Quick References::            
@end menu

@node    Tutorials for Languages, Internet Skills for Disconnected People, Projects, Projects
@section Tutorials for Languages

@node    Internet Skills for Disconnected People, Quick References, Tutorials for Languages, Projects
@section Internet Skills for Disconnected People

@node    Quick References,  , Internet Skills for Disconnected People, Projects
@section Quick References




@node    Two Halves, Concept Index, Projects, Top
@chapter Splitting Eev in Two Halves (and Their Frontier)

Eev is mainly composed of two parts: one that deals with creating and
following elisp hyperlinks, and another one that is concerned with
selecting some region of a text and sending it to an external program.
Usually the externel program is a shell, and ``sending'' the region
means just saving it into a certain file --- a ``temporary script''.

Actually this divison in two parts doesn't work very well --- the
frontier is blurry. In both halves what happens is that we are
executing something.

[Doing something means executing code.]

[When we follow a hyperlink with M-e the code is visible; when we
type, say, M-x eev, the code for eev is not visible, but it's not hard
to access, either. And the code that executes backward-sexp /
buffer-substring / read / eval is also not visible.]

[In a sense, M-x eev is just half of the execution of something; we
still need to go to a shell and make it do the rest. But it is also a
task complete in itself --- save the region somewhere.]

[Actually we are doing bigger tasks. What we are doing on the computer
is just a part of them.]



@menu
* Saving::                      Saving temporary scripts.
* Links::                       
* Actions to Code in Emacs::    
* Hybrid Functions::            
* More on Hyperlinks::          
@end menu

@node    Saving, Links, Two Halves, Two Halves
@section Saving Temporary Scripts

@example
(find-eimage0 (ee-eevfile "doc/diag-save.jpg") 1 0 'inline)
@end example

The diagram above shows what is involved in writing some lines of
code for an external program (most typically a shell) in Emacs and
then making the program execute those lines. I've made a commented
screenshot about that:





@example
(find-eevfile "doc/shot-f3.png")
@end example

The "technically simpler" way - that is, the one involving less
functions - to send commands to a shell is to write the commands as
text in a buffer, then mark the region containing these commands,
then type `M-x eev'; that will save the region in a certain
temporary file. The name of that file is given by a certain
environment variable - $EE, usually set to ~/eev-current/tmp/ee.sh -
and that value of variable is set in rcfiles - check your .bashrc,
.zshrc and .emacs after installation. Then to execute these commands
we go to a shell and type `ee'; `ee' is a shell function (also
defined in .bashrc/.zshrc) that "sources" the contents of $EE.

It is usually more practical to write the commands inside a "region"
bounded by delimiters (see the "shot-f3.png" screenshot above), then
just place the cursor inside that region and type <f3>. By default
<f3> will execute `eev-bounded', that will run `eev' in that bounded
region; but there are several ways to change this default. The
variable `eeb-defaults' contains the "default bounded function", the
delimiters to use, and other things; if you another "bounded
function" explicitly, say, with `M-x eelatex-bounded', that will set
the defaults - the next time you time <f3> it will run
`eelatex-bounded' using block delimiters that make more sense for
LaTeX - "\n%*\n" instead of "\n#*\n". So actually <f3> runs
`eeb-default'.

[Actually the diagram is a bit misleading... eev-bounded calls
eeb-default[s], or something like that. Explain and show how the old
implementation was]

@example
(find-eev "eev-dev.el")
@end example

@node    Links, Actions to Code in Emacs, Saving, Two Halves
@section The other half: following and creating Elisp hyperlinks

@example
(find-eimage0 "doc/diag-links.jpg" 1 2 'inline)
@end example

Emacs's support for executing Lisp expressions
(a.k.a. "S-expressions", or "sexps") embedded in any kind of text
lets us put some simple types of hyperlinks anywhere; for example,
typing `C-x C-f /usr/share/doc/bash' or `M-x find-file
/usr/share/doc/bash' has the same effect as placing the cursor after
the text (find-file "/usr/share/doc/bash") and executing that with
C-x C-e or with M-e; the effect is to create a buffer visiting
/usr/share/doc/bash/ - that, incidentally, is a directory, so the
buffer will be in "dired mode".

Eev defines the function `find-fline', that is a bit more adequate
than `find-file' for hyperlinks. For example,

@example
(find-fline "/usr/share/doc/bash/" "FAQ")
@end example

will do the same but will also place the cursor after the first
occurence of the string "FAQ" in the buffer. Most hyperlinks
functions in eev support this kind of finer control; the optional
arguments after the file name are called the "pos-spec-list".

If we need to create links to several files inside a certain
directory we can use `code-c-d': for example, (code-c-d "ud"
"/usr/share/doc/") will define a function `find-udfile' and several
others with names like `find-ud___' `ee-ud___'.

Until some time ago we had to write almost all these hyperlinks by
hand - there were a function called `inn', that would insert a
hyperlink to the last Info node visited, and `dff' and `dfa', that
would convert lines containing names of Debian packages into
hyperlinks to their descriptions, to their lists of files and to
their documentation directory, but just that.  Then in November of
2004 I got an idea for a group of functions that would take either a
file being visited, or a name of an Emacs Lisp function, or a key
sequence, or several other things, and would create a temporary
buffer containing several interesting hyperlinks about that thing.
For example: typing `M-h f' on a buffer visiting
/usr/share/doc/bash/examples/ gives us a buffer like this:

@example
 ___________________________________________________________ 
|(find-file-links "/usr/share/doc/bash/examples/")          |
|                                                           |
|(find-fline "/usr/share/doc/bash/examples/")               |
|                                                           |
|(find-bashfile "examples/")                                |
|(find-udfile "bash/examples/")                             |
|                                                           |
|                                                           |
|                                                           |
|--:**  *Elisp hyperlinks*   All L9      (Fundamental)------|
|___________________________________________________________|
@end example


In this case there aren't any surprising hyperlinks (but try @key{M-h
M-k <del>}, @key{M-h M-v ctl-x-map} or @key{M-h M-d bash}!) but it is
clear that we can not only follow each of these hyperlinks but also
edit them and copy them to more permanent places.

There are a few special keys for working on these hyperlinks. @key{M-h
M-2} or @key{M-h 2} will duplicate the current line, @key{M-h M-y} ---
the name is to remind us of @key{C-y} --- will paste the last element
of the kill ring as a pos-spec; for example, if the last string we
killed (say, with @key{C-w}) was ``scripts'' then @key{M-h M-2 M-h
M-y} will transform:

@example
(find-udfile "bash/examples/")
@end example

into:

@example
(find-udfile "bash/examples/")
(find-udfile "bash/examples/" "scripts")
@end example

There are currently no special keys to help us build code-c-d
expressions --- but check this:

@example
(find-efile "hippie-exp.el")
(global-set-key [f11] 'hippie-expand)
@end example

then try to type part of a file name and use @key{f11} to complete it.

See (find-eev "eev-insert.el")


@node    Actions to Code in Emacs, Hybrid Functions, Links, Two Halves
@section Emacs's tools for transforming actions into code

[To be written. Mention debugging, C-h k, interactive calls, keymaps,
modes.]



@node    Hybrid Functions, More on Hyperlinks, Actions to Code in Emacs, Two Halves
@section The frontier: hybrid functions

[To be written.]

#*
# (eev-bounded)
wget --help

#*
# (eev "wget --help")
# (find-sh "wget --help")

#*
inkscape $EEVDIR/doc/diag-save.svg
#*
# (eev "inkscape $EEVDIR/doc/diag-save.svg &")

eev or find-sh0 links inside lisp comments
eev or find-sh  links inside shell comments

hybrid links will exist every time that we are learning a new program
and still don't know how to make links to it using more common-looking
links.


@node    More on Hyperlinks,  , Hybrid Functions, Two Halves
@section More on hyperlinks (plus garbage)

[Fix this, give examples]

[and so we don't need to quote portions of external texts inside an
e-script; instead, we can point directly to the original sources. Eev
defines many kinds of "hyperlinks" like these: for text files, for
info manuals, for manpages, for ps/pdf/dvi/html files, for the output
of shell commands, for Emacs's own source code in Lisp or in C, for
Debian packages, for images, and a few others.]

The two hyperlinks below are equivalent:

  (find-man "1 expect" "\nCOMMANDS" "\n       interact [string1 body1] ...")
  (find-expcommand "interact [string1 body1] ...")

The first one opens the manpage for the program Expect (like "man 1
expect"), then searches for the first ocurrence of a newline followed
by the string "COMMANDS" -- that's the beginning of the section of the
manpage that describes expect's commands --, then searches for the
first place after that where the string "interact [string1 body1] ..."
(preceded by a newline and exactly 8 spaces) occurs; that's the point
of the manpage where is the description of the command "interact".

the underlying idea (or "belief"?) is that we should be able to
"record" all kinds of textual interactions with the machine in
e-scripts, in a way that will let us "play back" these interactions
later - maybe with modifications - and maybe even share them with
other people.

Also: readability. Recording the way to open up an info page as a
series of keys is not very useful... (give an example). Idem for
mouseclicks. Show how commands in Gimp have textual names:

# (find-sh "cd ~/.gimp-2.0/; sort menurc")

E-scripts may be a more powerful way to communicate
actions and procedures than either finished programs or texts and
how-tos, which usually describe most of the steps using human
languages.










@cindex Sample index entry

@node    Concept Index,  , Two Halves, Top
@unnumbered Concept Index

@printindex cp
@contents
@bye

@c Local Variables:
@c coding:  raw-text-unix
@c modes:   (fundamental-mode emacs-lisp-mode texinfo-mode)
@c sentence-end-double-space: t
@c End: