[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [tlaplus] Functions in the standard libraries



What do you mean that the algorithm is "wrong"?

On Saturday, December 17, 2022 at 8:35:00 AM UTC-5 radhakris...@xxxxxxxxx wrote:
I am able to use 

struct == [a |-> 97, b |-> 98, c |-> 99]

and a small subset to complete the model check phase. States are less in number now. But the code is very convoluted.

I believe you meant that this is what abstraction means ? The Pluscal code will be quite complex if I try to code the entire program

which isn't our goal. But. I have a question. If the Pluscal code isn't complete(in this case), the algorithm is wrong and the model

checker still completes. What does it mean for the specification ? How do I interpret that state ?

Thanks,

Mohan


On Friday, December 16, 2022 at 4:13:01 PM UTC+5:30 Stephan Merz wrote:
Please see https://github.com/tlaplus/tlaplus/issues/512.

But I believe that for your application, using an abstract set of characters is preferable.

Stephan

On 16 Dec 2022, at 11:11, mohan radhakrishnan <radhakris...@xxxxxxxxx> wrote:

Hello,
             I experimented with the changes and they work but the function

Ascii(char) == 96 + CHOOSE z \in 1 .. 26 :"abcdefghijklmnopqrstuvwxyz"[z] = char

throws this. I have converted the code in the book to Pluscal. Is that wrong ?

The exception was a java.lang.RuntimeException

: A non-function (a string) was applied as a function.


Thanks,
Mohan

On Thursday, December 15, 2022 at 1:30:33 PM UTC+5:30 Stephan Merz wrote:
Hello,

please check Section 16.1.10 of Specifying Systems, which discusses strings and characters and even contains a function to convert lower-case letters to ASCII. In short, TLA+ has no hardwired definition of what characters are. In your representation, the variables in_pattern and in_text are sequences of strings, not characters.

As far as I can tell, the algorithm is independent of any concrete representation of characters. I suggest that you make the set of characters as well as the inputs to the algorithm parameters of your specification, i.e. introduce (at the top of the module)

CONSTANT Character, text, pattern

ASSUME text \in Seq(Character) /\ pattern \in Seq(Character)

When you use TLC to check your algorithm, you can instantiate these constants by concrete values, e.g.

Character <- {a,b,c}  \* a set of model values
text <- << a, a, b, a, c >>
pattern <- << b, a >>

In this way, you'll also alleviate the effect of state explosion. If you want to go further than testing the algorithm on concrete inputs, you can replace the parameters `text' and `pattern' by variables and nondeterministically assign them finite (bounded-length) sequences of characters by writing something like

define {
  \* sequences of characters of length at most n
  CharSeq(n) == UNION { [1 .. k -> Character] : k \in 0 .. n }
}

[...]

variables text \in CharSeq(5), pattern \in CharSeq(3);

[...]

In this way, TLC will check the algorithm for all text / pattern strings of bounded length.

Remember that TLA+ is a language for specifying algorithms at a high level of abstraction, not a programming language. 

Hope this helps,
Stephan


On 15 Dec 2022, at 05:53, mohan radhakrishnan <radhakris...@xxxxxxxxx> wrote:

Hello,
           I was looking for functions to get the ASCII value of characters but realized I couldn't
find a list of all functions available. Should I look for a facility to add my own functions ?

Can I check if the code returns the correct value before checking the model and state spaces ? The state space for this is enormous anyway.

Thanks,
Mohan

-- 
You received this message because you are subscribed to the Google Groups "tlaplus" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tlaplus+u...@xxxxxxxxxxxxxxxx.
To view this discussion on the web visit https://groups.google.com/d/msgid/tlaplus/6c72f651-82f0-416c-b01f-744a7bef5662n%40googlegroups.com.


-- 
You received this message because you are subscribed to the Google Groups "tlaplus" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tlaplus+u...@xxxxxxxxxxxxxxxx.

--
You received this message because you are subscribed to the Google Groups "tlaplus" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tlaplus+unsubscribe@xxxxxxxxxxxxxxxx.
To view this discussion on the web visit https://groups.google.com/d/msgid/tlaplus/405d1a39-1eca-451c-b85e-f11aa8cf5b41n%40googlegroups.com.