And what is it, if it isn't?
Everything I've read about TCL states that everything is just a string in it. There can be some other types and structures inside of an interpreter (for performance), but at TCL language level everything must behave just like a string. Or am I wrong?
I'm using an IDE for FPGA programming called Vivado. TCL automation is actively used there. (TCL version is still 8.5, if it helps)
Vivado's TCL scripts rely on some kind of "object oriented" system. Web search doesn't show any traces of this system elsewhere.
In this system objects are usually obtained from internal database with "get_*" commands. I can manipulate properties of these objects with commands like get_property, set_property, report_property, etc.
But these objects seem to be something more than just a string.
I'll try to illustrate:
> set vcu [get_bd_cells /vcu_0]
/vcu_0
> puts "|$vcu|"
|/vcu_0|
> report_property $vcu
Property Type Read-only Value
CLASS string true bd_cell
CONFIG.AXI_DEC_BASE0 string false 0
<...>
> report_property "$vcu"
Property Type Read-only Value
CLASS string true bd_cell
CONFIG.AXI_DEC_BASE0 string false 0
<...>
But:
> report_property "/vcu_0"
ERROR: [Common 17-58] '/vcu_0' is not a valid first class Tcl object.
> report_property {/vcu_0}
ERROR: [Common 17-58] '/vcu_0' is not a valid first class Tcl object.
> report_property /vcu_0
ERROR: [Common 17-58] '/vcu_0' is not a valid first class Tcl object.
> puts |$vcu|
|/vcu_0|
> report_property [string range $vcu 0 end]
ERROR: [Common 17-58] '/vcu_0' is not a valid first class Tcl object.
So, my question is: what exactly is this "valid first class Tcl object"?
Clarification:
This question might seem like asking for help with Vivado scripting, but it is not. (I was even in doubt about adding [vivado] to tags.)
I can just live and script with these mystic objects.
But it would be quite useful (for me, and maybe for others) to better understand their inner workings.
Is this "object system" a dirty hack? Or is it a perfectly valid TCL usage?
If it's valid, where can I read about it?
If it is a hack, how is it (or can it be) implemented? Where exactly does string end and object starts?
Related:
A part of this answer can be considered as an opinion in favor of the "hack" version, but it is quite shallow in a sense of my question.
A first class Tcl value is a sequence of characters, where those characters are drawn from the Basic Multilingual Plane of the Unicode specification. (We're going to relax that BMP restriction in a future version, but that's not yet in a version we'd recommend for use.) All other values are logically considered to be subtypes of that. For example, binary strings have the characters come from the range [U+000000, U+0000FF], and integers are ASCII digit sequences possibly preceded by a small number of prefixes (e.g., - for a negative number).
In terms of implementation, there's more going on. For example, integers are usually implemented using 64-bit binary values in the endianness that your system uses (but can be expanded to bignums when required) inside a value boxing mechanism, and the string version of the value is generated on demand and cached while the integer value doesn't change. Floating point numbers are IEEE double-precision floats. Lists are internally implemented as an array of values (with smartness for handling allocation). Dictionaries are hash tables with linked lists hanging off each of the hash buckets. And so on. THESE ARE ALL IMPLEMENTATION DETAILS! As a programmer, you can and should typically ignore them totally. What you need to know is that if two values are the same, they will have the same string, and if they have the same string, they are the same in the other interpretation. (Values with different strings can also be equal for other reasons: for example, 0xFF is numerically equal to 255 — hex vs decimal — but they are not string equal. Tcl's true natural equality is string equality.)
True mutable entities are typically represented as named objects: only the name is a Tcl value. This is how Tcl's procedures, classes, I/O system, etc. all work. You can invoke operations on them, but you can only see inside to a limited extent.
Vivado TCL is not TCL. Vivado will not really document their language they call TCL, but refer you to the real TCL language documentation. Where Vivado TCL and TCL differ, you are left on your own without help. TCL was a poor choice for a scripting language given the very large data bases, so they had to bastardize it to get it half functional. You are better off getting help on the Xilinx forums then in general TCL forums. Why they went with TCL rather than python is beyond anyone's comprehension.
Related
While finalizing my upcoming Raku Advent Calendar post on sigils, I decided to double-check my understanding of the type constraints that sigils create. The docs describe sigil type constraints with the table
below:
Based on this table (and my general understanding of how sigils and containers work), I strongly expected this code
my %percent-sigil is List = 1,2;
my #at-sigil is Map = :k<v>;
to throw an error.
Specifically, I expected that is List would attempt to bind the %-sigiled variable to a List, and that this would throw an X::TypeCheck::Binding error – the same error that my %h := 1,2 throws.
But it didn't error. The first line created a List that seemed perfectly ordinary in every way, other than the sigil on its variable. And the second created a seemingly normal Map. Neither of them secretly had Scalar intermediaries, at least as far as I could tell with VAR and similar introspection.
I took a very quick look at the World.nqp source code, and it seems at least plausible that discarding the % type constraint with is List is intended behavior.
So, is this behavior correct/intended? If so, why? And how does that fit in with the type constraints and other guarantees that sigils typically provide?
(I have to admit, seeing an %-sigiled variable that doesn't support Associative indexing kind of shocked me…)
I think this is a grey area, somewhere between DIHWIDT (Docter, It Hurts When I Do This) and an oversight in implementation.
Thing is, you can create your own class and use that in the is trait. Basically, that overrides the type with which the object will be created from the default Hash (for %) and Array (for # sigils). As long as you provide the interface methods, it (currently) works. For example:
class Foo {
method AT-KEY($) { 42 }
}
my %h is Foo;
say %h<a>; # 42
However, if you want to pass such an object as an argument to a sub with a % sigil in the signature, it will fail because the class did not consume the Associatve role:
sub bar(%) { 666 }
say bar(%h);
===SORRY!=== Error while compiling -e
Calling bar(A) will never work with declared signature (%)
I'm not sure why the test for Associative (for the % sigil) and Positional (for #) is not enforced at compile time with the is trait. I would assume it was an oversight, maybe something to be fixed in 6.e.
Quoting the Parameters and arguments section of the S06 specification/speculation document about the related issue of binding arguments to routine parameters:
Array and hash parameters are simply bound "as is". (Conjectural: future versions ... may do static analysis and forbid assignments to array and hash parameters that can be caught by it. This will, however, only happen with the appropriate use declaration to opt in to that language version.)
Sure enough the Rakudo compiler implemented some rudimentary static analysis (in its AOT compilation optimization pass) that normally (but see footnote 3 in this SO answer) insists on binding # routine parameters to values that do the Positional role and % ones to Associatives.
I think this was the case from the first official Raku supporting release of Rakudo, in 2016, but regardless, I'm pretty sure the "appropriate use declaration" is any language version declaration, including none. If your/our druthers are static typing for the win for # and % sigils, and I think they are, then that's presumably very appropriate!
Another source is the IRC logs. A quick search quickly got me nothing.
Hmm. Let's check the blame for the above verbiage so I can find when it was last updated and maybe spot contemporaneous IRC discussion. Oooh.
That is an extraordinary read.
"oversight" isn't the right word.
I don't have time tonight to search the IRC logs to see what led up to that commit, but I daresay it's interesting. The previous text was talking about a PL design I really liked the sound of in terms of immutability, such that code could become increasingly immutable by simply swapping out one kind of scalar container for another. Very nice! But reality is important, and Jonathan switched the verbiage to the implementation reality. The switch toward static typing certainty is welcome, but has it seriously harmed the performance and immutability options? I don't know. Time for me to go to sleep and head off for seasonal family visits. Happy holidays...
What's equivalent of std::vector, std::deque and std::map in ObjectPascal (FreePascal compiler)?
In brief:
(vector) is an auto-resizing contiguous array
(deque) is an auto-sizing hybrid array of arrays giving near O(1) random
access while allowing O(1) push/pop from either end
(map, unordered_map) is an associative array
In general it is not logical to assume there are direct substitutes in some different language.
Currently FPC generics are a mix of old school C++ like generics (based on token replay), and Delphi more .NET styled generics (fully declarative, but more limited for value types for languages without autoboxing).
Anyway, I'll give it a try:
TList or its generic variants. (TList<> in Delphi or fgl.Tf*List in unit fgl)
No standard type, I have an array of array generic class, but that is optimized to avoid some of the problems of ordered Lists (insertion performance) while still being an ordered type. I've put it on http://www.stack.nl/~marcov/genlight.pas, maybe it gives you some ideas on how to approach the problem.
None yet. TDictionary once http://bugs.freepascal.org/view.php?id=27206 is committed. Currently usually TAVLTree is used.
There is also some generics including a simple deque in packages/fcl-stl, I suggest you check it out.
This is normal definition of some function as I know:
real function f(x)
real x
f = (sin(x))**2*exp(-x)
end function f
But I want to define a function from some string, for example the program will ask me to write it, and then it will define the function f in a program. Is this possible in Fortran?
What you are looking for is possible in reflective programming languages, and is not possible in Fortran.
Quote from the link above:
A language supporting reflection provides a number of features available at runtime that would otherwise be very obscure to accomplish in a lower-level language. Some of these features are the abilities to:
Discover and modify source code constructions (such as code blocks, classes, methods, protocols, etc.) as a first-class object at runtime.
Convert a string matching the symbolic name of a class or function into a reference to or invocation of that class or function.
Evaluate a string as if it were a source code statement at runtime.
Create a new interpreter for the language's bytecode to give a new meaning or purpose for a programming construct.
I worked on a project once that tried to achieve something similar. We read in a string that contained a string with named variables and mathematical operations (a function if you will). In this string the variables then got replaced by their numerical values and the terms were evaluated.
The basic idea is not to too difficult, but it requires a lot of string manipulations - and it is not a function in the context of a programming language.
We did it like this:
Recursively divide the string at +,-,/,*, but remember to honor brackets
If this is not possible (without violating bracketing), evaluate the remaining string:
Does it contain a mathematical expression like cos? Yes => recurse into arguments
No => evaluate the mathematical expression (no variables allowed, but they got replaced)
This works quite well, but it requires:
Splitting strings
Matching in strings
Replacing strings with other strings, etc.
This is not trivial to do in Fortran, so if you have other options (like calling an external tool/script that returns the value), I would look into that - especially if you are new to Fortran!
I am defining a new TCL command whose implementation is C++. The command is to query a data stream and the syntax is something like this:
mycmd <arg1> <arg2> ...
The idea is this command takes a list of arguments and returns a list which has the corresponding data for each argument.
My colleague commented that it is best just to use a single argument and when multi values are needed, just call the command multiple times.
There are some other discussions, but one thing we cannot agree with each other is, the performance.
I think my version, list of argument should be quicker because when we want multi arguments, it is one time cost going through TCL interpreter.
His comment is new to me -
function implementation is cached
accessing TCL function is quicker than accessing TCL data
Is this reasoning sound?
If you use Tcl_EvalObjv to invoke the command, you won't go through the Tcl interpreter. The cost will be one hash-table lookup (or less, if you reuse the Tcl_Obj* containing the command name) and then you'll be in the implementation of the command. Otherwise, constructing a list Tcl_Obj* (e.g., with Tcl_NewListObj) and then calling Tcl_EvalObj is nearly as cheap, as that's a special case because the list construction code is guaranteed to produce lists that are also substitution-free commands.
Building a normal string and passing that through Tcl_Eval (or Tcl_EvalObj) is significantly slower, as that has to be parsed. (OTOH, passing the same Tcl_Obj* through Tcl_EvalObj multiple times in a row will be faster as it will be compiled internally to bytecode.)
Accessing into values (i.e., into Tcl_Obj* references) is pretty fast, provided the internal representation of those values matches the type that the access function requires. If there's a mismatch, an internal type conversion function may be called and they're often relatively expensive. To understand internal representations, here's a few for you to think about:
string — array of unicode characters
integer — a C long (except when you spill over into arbitrary precision work)
list — array of Tcl_Obj* references
dict — hash table that maps Tcl_Obj* to Tcl_Obj*
script — bytecoded version
command — pointer to the implementation function
OK, those aren't the exact types (there's often other bookkeeping data too) but they're what you should think of as the model.
As to “which is fastest”, the only sane way to answer the question is to try it and see which is fastest for real: the answer will depend on too many factors for anyone without the actual code to predict it. If you're calling from Tcl, the time command is perfect for this sort of performance analysis work (it is what it is designed for). If you're calling from C or C++, use that language's performance measurement idioms (which I don't know, but would search Stack Overflow for).
Myself? I advise writing the API to be as clear and clean as possible. Describe the actual operations, and don't distort everything to try to squeeze an extra 0.01% of performance out.
I have this question in a homework assignment for my Computer Languages class. I'm trying to figure out what each one means, but I'm getting stuck.
Errors in a computer program can be
classified according to when they are
detected and, if they are detected at
compile time, what part of the
compiler detects them. Using your
favorite programming language, give an
example of:
(a) A lexical error, detected by the
scanner.
(b) A syntax error, detected by the
parser.
(c) A static semantic error, detected
(at compile-time) by semantic
analysis.
(d) A dynamic semantic error, detected
(at run-time) by code generated by the
compiler.
For (a), I think this is would be correct: int char foo;
For (b), int foo (no semicolon)
For (c) and (d), I'm not sure what is being asked.
Thanks for the help.
I think it's important to understand what a scanner is, what a parser is and how they are involved in the compilation process.
(I'll try my best at a high-level explanation)
The scanner takes a sequence of characters (a source file) and converts it to a sequence of tokens. e.g., sees the text if 234 ) and converts to the tokens, IF INTEGER RPAREN (there's more to it but should be enough for the example).
Another way you can think of how the scanner works is that it takes the text and makes sure you use the correct keywords and not makes them up. It has to be able to convert the entire source file to the associated language's recognized tokens and this varies from language to language. In other words, "Does every piece of text correspond to a construct a language understands". Or better put with an example, "Do all these words found in a book, belong to the English language?"
The parser takes a sequence of tokens (usually from the scanner) and (among other things) sees if it is well formed. e.g., a C variable declaration is in the form Type Identifier SEMICOLON.
The parser checks "Does this sequence of tokens in this order make sense to me?" And similarly the analogy, "Does this sequence of English words (with punctuation) form complete sentences?"
C asks for errors that can be found when compiling the program. D asks for errors that you see when running the program after it compiled successfully. You should be able to distinguish these two by now hopefully.
I hope this helps you get a better understanding and make answering these easier.
I'll give it a shot. Here's what I think:
a. int foo+; (foo+ is an invalid identifier because + is not a valid char in identifiers)
b. foo int; (Syntax error is any error where the syntax is invalid - either due to misplacement of words, bad spelling, missing semicolons etc.)
c. Static semantic error are logical errors. for e.g passing float as index of an array - arr[1.5] should be a SSE.
d. I think exceptions like NullReferenceException might be an example of DME. Not completely sure but in covariant returns that raise an exception at compile time (in some languages) might also come in this category. Also, passing the wrong type of object in another object (like passing a Cat in a Person object at runtime might qualify for DME.) Simplest example would be trying to access an index that is out of bounds of the array.
Hope this helps.