curiosity with here document.

curiosity with here document.

[Ray Andrews]

I often comment out blocks of code with here-document, and I just found 
this strange thing:

    /test1 ()/
    /{/
    /    if [[ "$1" = 'howdy' ]];/

    /: <<'COMMENT'/
    /blah blah
    COMMENT/

    /    then/
    /    echo stranger/
    /    fi/
    /}/

    /$ test1 no/
    /stranger/

There's no issue if I put the 'then' above the here-document. How should 
I understand that?  It is as if the here-document is hijacking the 'if' 
test and returning true. I'd expect the 'if' to go hunting for it's 
'then' and if it doesn't find it, then throw and error, but not let the 
here-document interlope.  But if this is proper then it's a caution as 
to using these things for commenting.

Re: curiosity with here document.

[ZyX]

26.11.2015, 21:38, "Ray Andrews" <rayandrews@eastlink.ca>:
> I often comment out blocks of code with here-document, and I just found
> this strange thing:
>
>     /test1 ()/
>     /{/
>     / if [[ "$1" = 'howdy' ]];/
>
>     /: <<'COMMENT'/
>     /blah blah
>     COMMENT/
>
>     / then/
>     / echo stranger/
>     / fi/
>     /}/
>
>     /$ test1 no/
>     /stranger/
>
> There's no issue if I put the 'then' above the here-document. How should
> I understand that? It is as if the here-document is hijacking the 'if'
> test and returning true. I'd expect the 'if' to go hunting for it's
> 'then' and if it doesn't find it, then throw and error, but not let the
> here-document interlope. But if this is proper then it's a caution as
> to using these things for commenting.

This has nothing to do with here document. You can have any *sequence* of commands as the `if` condition, including `:`. What this sequence returns is used for the if: 

    if true; false
    then
        echo True
    else
        echo False
    fi

will echo False.

Re: curiosity with here document.

[ZyX]

26.11.2015, 21:38, "Ray Andrews" <rayandrews@eastlink.ca>:
> I often comment out blocks of code with here-document, and I just found
> this strange thing:
>
>     /test1 ()/
>     /{/
>     / if [[ "$1" = 'howdy' ]];/
>
>     /: <<'COMMENT'/
>     /blah blah
>     COMMENT/
>
>     / then/
>     / echo stranger/
>     / fi/
>     /}/
>
>     /$ test1 no/
>     /stranger/
>
> There's no issue if I put the 'then' above the here-document. How should
> I understand that? It is as if the here-document is hijacking the 'if'
> test and returning true. I'd expect the 'if' to go hunting for it's
> 'then' and if it doesn't find it, then throw and error, but not let the
> here-document interlope. But if this is proper then it's a caution as
> to using these things for commenting.

And when writing such questions do read documentation. The specification for `if` looks like

    if list then list [ elif list then list ] ... [ else list ] fi

and using “list” for *both* `if` condition and `then` block should have given you an answer.

Re: curiosity with here document.

[Ray Andrews]

On 11/26/2015 11:03 AM, ZyX wrote:
> 26.11.2015, 21:38, "Ray Andrews" <rayandrews@eastlink.ca>:
>> And when writing such questions do read documentation. The specification for `if` looks like
>>
>>      if list then list
Until you know what 'list' entails, " if list then list " doesn't answer 
every question.  I've never seen anything like this previously so I 
wanted some knowledgeable comment, begging your pardon and thanks, I now 
know that a here-doc is true as far as any following 'then' is 
concerned, and any preceding 'if' test is ignored, which seems strange 
to me, but there it is.  So using that for commenting is not always 
safe; lesson learned.

Re: curiosity with here document.

[Bart Schaefer]

On Nov 26,  3:19pm, Ray Andrews wrote:
}
} lesson learned.

Except it's the wrong lesson!  What you *ought* to learn is, first, that
for purposes of determining the "truth value" (zero or nonzero return
status),

    if command1; command2; command3; then ...

is the same as

    function three_commands {
	command1; command2; command3
    }
    if three_commands; then ...

And second, that [[ ... ]] is just a command like any other command,
it's not magically connected to "if".  Further, ":" is also a command
like any other command.

And third, that a here-document is just the standard input of the
command it follows, so

    : <<END
    some stuff
    END

is (again for purposes of determining truth value) the same as

    :

and the colon command is "true" for purposes of if/while/until/etc.,
it has a return status of zero.  So what you've written is

    if [[ ... ]]; : ; then ...

and colon is true and is the last command before the "then", so the
true branch is taken.

The here-document itself has no true or false interpretation, and the
presence of the here-document is not what is causing "if" to behave
the way it does.

Re: curiosity with here document.

[Ray Andrews]

On 11/26/2015 05:56 PM, Bart Schaefer wrote:
> On Nov 26,  3:19pm, Ray Andrews wrote:
> }
> } lesson learned.
>
> Except it's the wrong lesson!
Well no, it is a lesson to be careful trying to use here-docs to comment 
blocks. As to why it goes off the rails, I was close to understanding it 
anyway--I'd expect legal code like that to have some sort of truth 
value, inconvenient as it may be.  I'm not really surprised by the 
correct understanding.
>      function three_commands {
> 	command1; command2; command3
>      }
>      if three_commands; then ...
Sure, but in practice I don't think we see that sort of 'list' after an 
'if' very often.  It seems that only the truth value of the last 
statement is acted upon as far as any 'if-then' structure, so as I said, 
the original 'if' seems to be orphaned.  So what?  It's me put the 
parser into that strange position, I'm not bitchin', it did what it had 
to do. Don't want orphan tests?  Don't make them.
> And second, that [[ ... ]] is just a command like any other command,
> it's not magically connected to "if".

That's the sort of fact that one might think one understands without 
really understanding it.  I probably can't hardly help C-ing it the C 
way.  I appreciate these deep corrections.
> And third, that a here-document is just the standard input of the
> command it follows, so
>
>      : <<END
>      some stuff
>      END
>
> is (again for purposes of determining truth value) the same as
>
>      :

Yeah, I'm really not as thick as I seem about this.  Wanting the 
'comment' to be available at any point wasn't realistic. Not one of my 
better questions.

BTW, I wish there was a way of going back to the archives and deleting 
lousy questions, I'd sure cut out a pile of mine. Library of Babel.