How to avoid function invocation in COMPOSE-like circumstances? - function

In the following code, we can assign the result of a GET-WORD to p through a SET-WORD, and then use it under the new name:
p: :print
p [{Hello} {World}]
But what if you are using COMPOSE, and you find a situation such as this?
do compose [p: (:print)]
p [{Hello} {World}]
That gives an error:
*** ERROR
** Script error: -unnamed- is missing its value argument
** Where: do
** Near: do compose [p: (:print)] p ["Hello" "World"]
So it's like function values in a block are "live" when seen in the interpreter...whether they were fetched as an evaluative result or not. (It would seem they should be inert unless fetched or applied somehow, otherwise such assignments are not possible from within a COMPOSE or similar.)
It seems you have to quote a get-word, such as:
do compose [p: (quote :print)]
p [{Hello} {World}]
That could do the trick to make p the print function. But can you do it without going through a GET-WORD or similar as a proxy?

Yes, you can "disarm" the active function! value with a feature of the DO-dialect:
>> do probe compose [p: quote (:print)]
[p: quote make native! [[
"Outputs a value followed by a line break."
value [any-type!] "The value to print"
]]]
>> p [{Hello} {World}]
Hello World
They key here is the special argument passing mode used for the single argument of QUOTE:
>> source quote
quote: make function! [[
"Returns the value passed to it without evaluation."
:value [any-type!]
][
:value
]]
This argument passing mode, unimaginatively called "get arguments", inhibits evaluation of the argument value. So in our particular case, it prevents the "active" nature of the function! value to come to bear.
For more details about argument passing modes, you might want to have a look at this recent treatise about literal and get arguments, which compares the differences between Rebol 2 and Rebol 3 to give a historical perspective.

Use the following way to create the get-word, it buys you an additional level of indirection:
>> do compose [p: (to-get-word 'print)]
>> p [{Hello} {World}]
Hello World
If you want the assignment to be done using a DO [...] form, then you need a word assigned to the anonymous function in order to manipulate it in a passive way.
An alternative option is to do the assignment outside of the composed block:
p: first compose [(:print)]

The first question is what do you need this for?
Because it works without compose.
Do [p: :print]
P 1
And without a get-word
Do [p: get first [print]]
Edit:
I don't understand the implications for the C++ binding, but 'compose is used to evaluate parts of a block selectively. In your example you want to evaluate a part, but still want it not being evaluated, so you need to stop it from being evaluated too early, for example:
do compose [p1: (to get-word! 'x)]
If you want to use compose, and no get-/lit-word!, you could try quote:
do compose [p4: get quote ( quote print)]
The inner 'quote guards 'print from being evaluated during compose, the outer one guards durng the 'do, so that 'get can get the value.

Related

Use of brackets and space character in Tcl

I am still confused about the usage of the bracket i.e () [] and {} use in Tcl. I always get caught out using the wrong bracket, having missed brackets when it was required to use them or having used too many of them. Besides this, I am also getting confused by Tcl giving me different result depending on presence or absence of space character (in math expression) and also if I have used more than one space character in succession.
Can someone please give me the basic rules that I must keep in mind to get out of this mess. Brackets have always been simple to use in C and some other languages but here they are totally different.
At the level you're looking at, Tcl is very different to any other language you've ever worked with. The heart of Tcl is defined by the Tcl(n) manual page, which states that (among other things):
Whitespace separates words. Every command takes its arguments as a sequence of words. Newlines and semicolons separate command calls; they're totally equivalent, but good style is to use a newline instead of a semicolon.
{braces} are used mainly for quoting text so that it is passed to commands with no substitutions or word separation performed on it. They nest properly. Braces are also used after $ to do variable substitution in a few cases: that's a rare use.
"double quotes" are used for quoting text so that it is passed to commands with substitutions applied, but no word separation.
[brackets] are a command substitution. They are replaced with the result of running the script inside the bracket. The script is usually a single command.
(parentheses) only have one base language use: for (associative) array elements. Thus, $a(b) is a variable substitution that will use the value of the b element in the a array.
The rest of what people call Tcl is really just a standard library, a set of commands to get you started. Some are fundamental. For example:
if is a conditional command, evaluating a branch (a script) if a condition is true. In order for this to be meaningful, the branch has to be not evaluated until the condition has been evaluated and tested; that pretty much requires putting it in braces.
while is a looping command, and not only do you want to brace its body (that's probably going to be evaluated over and over) but you also want to put the condition expression in braces as well as you definitely want that to be reevaluated each time round the loop.
proc is a command that makes your own custom commands. The body of the procedure definitely is something you want to evaluate later; it goes in braces.
expr is a general expression evaluation command. Under all normal circumstances, you'll want to put its expression in braces so that the code can be compiled and won't have double substitution problems. Note that expressions often make heavy use of parentheses: they have additional meanings in expression syntax. In particular, apart from being array element lookups, they're also used for function calls and grouping.
Note that if and while also use that same expression evaluation engine. They just use the result of the expression to decide what to do.
Scoping is a matter for commands to decide. The usual commands for dealing with introducing a scope are proc and namespace eval. This is nothing like C, C++, Java, C#, or Javascript; they have different rules. Variables are local to their procedure unless you explicitly say otherwise.
The community practice is to do calls like this:
if { $foo(bar) > (17 + $grill) * 7 } {
# This is a comment; it lasts to the end of the line
puts "the foobar $foo(bar) is too large"
set foo(bar) [ComputeSmallerValue $grill]
}
That is, barewords (if and puts) are unquoted, expressions and inner scripts are brace-quoted, parentheses are used where meaningful but most for arrays and expressions, whitespace separates all words, inner scripts are indented (usually by 4) for clarity (it doesn't have semantic meaning, but it sure helps with reading), and “blocks” use egyptian braces so that you don't have to add backslashes all over the place.
You don't have to follow these rules (they're guidelines, not the law) but they make your life easier if you do. Sometimes you do need to break the rules, but then you should know to be careful.
You cannot compare Tcl to C. In C, {} defines scope. In Tcl, {} is a grouping operator.
In Tcl, {} may group a string:
{hello world}
Or a list:
{a b c d e f g h}
Or a script:
{
puts -nonewline {hello }
puts world\n
}
Every command is simply a series of groups (which may be a word, a list,
an expression or a script):
{if} {true} { puts "hello\n" }
Of course, you don't need to put braces around every word,
but you do need braces to enclose a script:
if true { puts hello\n }
Generally, for the if statement, not bracing the expression is a bad idea,
so this is better:
if { true } { puts hello\n }
This simple rule creates Tcl's remarkably simple syntax. Every command is simply
a series of groups, whether a word, an expression, a list or script:
if expr script
while expr script
proc name argument-list script
puts string
for initialization condition nextloop script
The one important thing to remember is whenever an expression is wanted, it
should be enclosed within braces in order to prevent early substitution. e.g.:
set i 0
while { $i < 10 } {
incr i
}
The square brackets, [], are replaced with the output of a command enclosed
by the square brackets:
set output [expr {2**5}]
Parentheses are used within expressions as usual:
set output [expr {(2**5)+2}]
And for arrays:
set i 0
while { $i < 5 } {
set output($i) [expr {2**$i}]
incr i
}
parray output

Printing variables including functions from Makefile and/or variable introspection

If you iterate over .VARIABLES and print each, any true variable can be printed correctly with the following rule:
print_variables: $(foreach V,$(.VARIABLES),print-$(V)) .phony_explicit
print-%: .phony_explicit; #echo "$* = \"$($*)\""
.PHONY: .phony_explicit ...
A 0- or 1-line function will still work, but any more will result in Syntax error: Unterminated quote string. Just one multiline function will break the entire print_variables rule. As a workaround, I have added ;\ to each line in my function definitions, but that won't fix existing multiline functions (either via includes from this makefile or via other makefiles including this one.) What can I do? Is there a container of just function variables, or a way to test if a variable is a function definition?
A simple minimal example would be easier to understand; this has nothing to do with .VARIABLES, pattern rules, etc. (and I'm not sure what the point of the .phony_explicit prereq is..)
define F
foo
bar
endef
print: ; echo "F = $(F)"
will show the problem:
echo "F = foo
/bin/sh: 1: Syntax error: Unterminated quoted string
This is because when make sees a variable that contains newlines in a recipe, it assumes that the newlines mean you want the lines of the variable to become lines in the recipe.
First in general you should use single-quotes around strings you send to the shell, unless you need the shell to expand them; it won't help in this situation but in general it's much safer.
There's no way to undo that, really. You have a number of options.
The first is to not use echo but instead use the make function info:
print-F: ; $(info F = "$(F)")
yields:
F = "foo
bar"
Another option is to use subst to replace the newlines with some other value. The new value cannot itself contain explicit newlines, but you can ask the shell to print a newline for you:
# Create a variable containing a single newline
# Note this must contain TWO newlines!
define NL
endef
print-F: printf 'F = "$(subst %,%%,$(subst $(NL),\n,$(F))"\n'
Yields:
printf 'F = "foo\nbar"\n'
F = "foo
bar"
One final option is to convert your makefile to use the .ONESHELL feature, but I assume that's a step too far just to get this debugging output available :).

tcl scripts, struggling with [...] and [expr ...]

I can't understand how assignments and use of variables work in Tcl.
Namely:
If I do something like
set a 5
set b 10
and I do
set c [$a + $b]
Following what internet says:
You obtain the results of a command by placing the command in square
brackets ([]). This is the functional equivalent of the back single
quote (`) in sh programming, or using the return value of a function
in C.
So my statement should set c to 15, right?
If yes, what's the difference with
set c [expr $a + $b]
?
If no, what does that statement do?
Tcl's a really strict language at its core; it always follows the rules. For your case, we can therefore analyse it like this:
set c [$a + $b]
That's three words, set (i.e., the standard “write to a variable” command), c, and what we get from evaluating the contents of the brackets in [$a + $b]. That in turn is a script formed by a single command invocation with another three words, the contents of the a variable (5), +, and the contents of the b variable (10). That the values look like numbers is irrelevant: the rules are the same in all cases.
Since you probably haven't got a command called 5, that will give you an error. On the other hand, if you did this beforehand:
proc 5 {x y} {
return "flarblegarble fleek"
}
then your script would “work”, writing some (clearly defined) utter nonsense words into the c variable. If you want to evaluate a somewhat mathematical expression, you use the expr command; that's it's one job in life, to concatenate all its arguments (with a space between them) and evaluate the result as an expression using the documented little expression language that it understands.
You virtually always want to put braces around the expression, FWIW.
There are other ways to make what you wrote do what you expect, but don't do them. They're slow. OTOH, if you are willing to put the + first, you can make stuff go fast with minimum interference:
# Get extra commands available for Lisp-like math...
namespace path ::tcl::mathop
set c [+ $a $b]
If you're not a fan of Lisp-style prefix math, use expr. It's what most Tcl programmers do, after all.
set c [$a + $b]
Running the above command, you will get invalid command name "5" error message.
For mathematical operations, we should rely on expr only as Tcl treats everything as string.
set c [expr $a + $b]
In this case, the value of a and b is passed and addition is performed.
Here, it is always safe and recommended to brace the expressions as,
set c [expr {$a+$b}]
To avoid any possible surprises in the evaluation.
Update 1 :
In Tcl, everything is based on commands. It can a user-defined proc or existing built-in commands such as lindex. Using a bare-word of string will trigger a command call. Similarly, usage of [ and ] will also trigger the same.
In your case, $a replaced with the value of the variable a and since they are enclosed within square brackets, it triggers command call and since there is no command with the name 5, you are getting the error.

Tcl: Is parameter evaluation guaranteed to be left-to-right?

I have a Tcl program where I often find expressions of the following kind:
proc func {} {...}
...
lappend arr([set v [func]]) $v
The intended meaning of the last line is
set v [func]
lappend arr($v) $v
It obviously works. What I would like to know: Does it work "by accident", or does Tcl guarantee, that the first parameter passed to lappend is evaluated before the second?
Tcl is always evaluated from left to right as you can read on the documentation, I quote the part:
Substitutions take place from left to right, and each substitution is evaluated completely before attempting to evaluate the next. Thus, a sequence like:
set y [set x 0][incr x][incr x]
will always set the variable y to the value, 012.
Agreed with Jerry. Adding some flavor in it.
Tcl commands are evaluated in two steps : parsing & execution.
First the Tcl interpreter parses the command string into words, performing substitutions along the way.
Then a command procedure processes the words to produce a result string. Each command has a separate command procedure.
Let us consider the following code.
%set input "The cat in the hat"
The cat in the hat
%string match "*at in*" $input
1
In the parsing step the Tcl interpreter applies the rules described in this chapter to divide the command up into words and perform substitutions.
Parsing is done in exactly the same way for every command. During the parsing step the Tcl interpreter does not apply any meaning to the values of the words. Tcl just performs a set of simple string operations such as replacing the characters $a with the string stored in variable a. Tcl does not know or care whether a or the resulting word is a number or the name of a widget or anything else.
In the execution step meaning is applied to the words of the command. Tcl treats the first word as a command name, checking to see if the command is defined and locating a command procedure to carry out its function. If the command is defined then the Tcl interpreter invokes its command procedure, passing all of the words of the command to the command procedure. The command procedure is free to interpret the words in any way that it pleases, and different commands apply very different meanings to their arguments.
Major rule to remember here
Tcl parses a command and makes substitutions in a single pass from left to right. Each character is scanned exactly once.
At most a single layer of substitution occurs for each character; the result of one substitution is not scanned for further
substitutions.
Reference : Tcl and the Tk Toolkit

Break on namespace function in gdb (llvm)

I'm trying to step through llvm's opt program (for an assignment) and the instructor suggested setting a breakpoint at runOnFunction. I see this in one of the files:
bool InstCombiner::runOnFunction(Function &F) { /* (Code removed for SO) */ }
but gdb does not seem to find the runOnFunction breakpoint. It occurred to me that the problem might be namespaces? I tried this but gdb never breaks, it just creates the fooOpt.s file:
(gdb) b runOnFunction
Function "runOnFunction" not defined.
Make breakpoint pending on future shared library load? (y or [n]) y
Breakpoint 1 (runOnFunction) pending.
(gdb) r -S -instcombine -debug -o ~/Desktop/fooOpt.s ~/Desktop/foo.s
I'm on a Mac so I don't have objdump but otool produces 5.6 million lines, wading through that for a starting point does not seem reasonable as runOnFunction appears more than once there.
Gdb has several builtin commands to find name of such functions. First is info functions, which can be used with optional regexp argument to grep all available functions, https://sourceware.org/gdb/current/onlinedocs/gdb/Symbols.html
info functions regexp
Print the names and data types of all defined functions whose names contain a match for regular expression regexp. Thus, ‘info fun step’ finds all functions whose names include step; ‘info fun ^step’ finds those whose names start with step. If a function name contains characters that conflict with the regular expression language (e.g. ‘operator*()’), they may be quoted with a backslash.
So, you can try info functions runOnFunction to get the name. Sometimes it can be useful to add quotes around name when doing break command.
The other way is to use rbreak command instead of break (b). rbreak will do regexp search in functions names and may define several breakpoints: https://sourceware.org/gdb/current/onlinedocs/gdb/Set-Breaks.html#Set-Breaks
rbreak regex
Set breakpoints on all functions matching the regular expression regex. This command sets an unconditional breakpoint on all matches, printing a list of all breakpoints it set. ...
The syntax of the regular expression is the standard one used with tools like grep. Note that this is different from the syntax used by shells, so for instance foo* matches all functions that include an fo followed by zero or more os. There is an implicit .* leading and trailing the regular expression you supply, so to match only functions that begin with foo, use ^foo.
(or even rbreak file:regex to limit search to single source file)
PS: if you want, you can turn on or off C++ function name demangling with set print demangle on or off (https://sourceware.org/gdb/current/onlinedocs/gdb/Debugging-C-Plus-Plus.html#Debugging-C-Plus-Plus). With demangling turned off it will be easier to copy function name to break command.