OpenEuphoria

1. value() problem maybe

• Posted by slowmaker Feb 25, 2013
• 1725 views

I think value() may be choking on back ticks.

If the error is mine, could someone please help me find it?

To me, if the value() routine does not choke on debug1 (below), then it should not choke on debug2. It should just evaluate the string objects to new string objects with the following character contents:

""" eval a triple-full-quote string """

` eval a back tick string `

However, only the first one works. As far as I can tell, any simple sequence starting with the back tick will make value() fail.

I have not tried get() yet.

sequence debug1 = `""" eval a triple-full-quote string """ ` 
sequence debug2 = """` eval a back tick string ` """ 
integer debug3 = '`' 
 
include std/io.e 
include std/get.e 
 
 
sequence evaluated = value(debug1, 1, GET_LONG_ANSWER) 
 
if evaluated[1] = GET_SUCCESS then 
	puts(1, "evaluated okay this time! Heres what it evaluated to:\n") 
	printf(1,"%s\n",{debug1}) 
else 
	printf(1,"DID NOT EVAL SUCCESSFULLY! Only got %d characters when evaluating this:\n%s\n",{evaluated[3], debug1})	 
end if 
 
puts(1,"\n") 
 
evaluated = value(debug2, 1, GET_LONG_ANSWER) 
if evaluated[1] = GET_SUCCESS then 
	puts(1, "evaluated okay this time! Heres what it evaluated to:\n") 
	printf(1,"%s\n",{debug2}) 
else 
	printf(1,"DID NOT EVAL SUCCESSFULLY! Only got %d characters when evaluating this:\n%s\n",{evaluated[3], debug2})	 
end if 
 
puts(1,"\n") 
evaluated = value({debug3}, 1, GET_LONG_ANSWER) 
if evaluated[1] = GET_SUCCESS then 
	puts(1, "evaluated okay this time! Heres what it evaluated to:\n") 
	printf(1,"%s\n",{debug3}) 
else 
	printf(1,"DID NOT EVAL SUCCESSFULLY! Only got %d characters when evaluating this:\n%s\n",{evaluated[3], debug3})	 
end if 
puts(1,"\n") 

3. Re: value() problem maybe

• Posted by ArthurCrump Feb 25, 2013
• 1639 views

See also ticket:638

5. Re: value() problem maybe

• Posted by mattlewis (admin) Feb 26, 2013
• 1559 views

I just pushed up a change to add multi-line string support. You can download the updated std/get.e here.

It doesn't cover everything in the ticket, since it doesn't handle new comment styles or dollar terminators.

Matt

7. Re: value() problem maybe

• Posted by mattlewis (admin) Feb 26, 2013
• 1540 views

According to the documentation it ignores comments like whitespace. There is a local function, read_comment() in std/get.e, but it only deals with line comments. It's been there since at least 2008.

Matt

9. Re: value() problem maybe

• Posted by slowmaker Apr 10, 2013
• 1429 views

Okay, here's the start of my attempt: http://openeuphoria.org/pastey/209.wc (do we have a way to attach files? not sure pastey is the appropriate thing for me to do here)

I took a swing at updating get.e to match the documentation. Here are the changes in this version:

    modified char() type range 
 
    in get_ch(), changed: 
	if ch = GET_EOF then 
    to 
        if ch != GET_EOF then 
 
    in skip_blanks(), changed to a function returning count of whitespace skipped 
 
    changed Get2() to get_verbose() and made it mostly a call to Get() with only  
    the extra data manipulation occuring in get_verbose() 
 
    added support for: 
       sequence terminator $ 
       new escape sequences \e, \E, \0, \b, \x, \u, \U 
       raw strings (built around Matt Lewis's get_heredoc()) """ """, ` ` 
       hex string specifiers x"FEED" 
       binary string specifiers b"BEEF" 
       ignoring block comments /* */ 
       underline spacers in numbers  123_999_000 
       binary number notation 0b10101010 
       hex number notation 0xBEADFADE 
       octal number notation 0t01234567 
       decimal number notation 0d9876543210 
     
    various other changes made necessary and/or convenient by the changes listed above;  
    new support functions, miscellaneous tweaks made while bug chasing (WinMerge, diff,  
    etc. will make your life easier there). 
 

I also made one change (mistakenly) that does not match the docs but which I decided to keep and make into a feature request http://openeuphoria.org/ticket/865.wc:

This version does not require whitepace between Euphoria objects, except where the nature of the object automatically enforces it (e.g. if you butt two numbers together, obviously they will be interpreted as one, which you presumably don't want; neither this version or the old one is telepathic, can't foresee that...).

This was originally simple forgetfulness. I didn't remember that requirement from the docs, and just assumed it should be able to handle stuff like 1234"string" as two consecutive objects. By the time I noticed that part of the docs, I already had it working, so I am requesting that everybody mull that one over as a possible official behavior change (ticket should hopefully be entered by the time you read this).

So...bug reports definitely welcome; I have only done rudimentary bug testing. I will try to attend to them when and as I can, just bear in mind that my screen name is fairly accurate.