Conference noted::hackers_v1

I've been known to search the parameter for lowercase and embedded
spaces to decide whether it needs quotes added back on.  Adding the
quotes is easy: MAIL/SUBJ="""''P2'"""

What I really need is a way to keep quotes embedded in a parameter 
from screwing up its use.  That is, if the user passes the string
"foo ""bar""", P2 is {foo "bar"} and the above turns into 
{MAIL/SUBJ="foo "bar""}, and mail bitches because it tried to parse 'bar'.

It would be a lot simpler if DCL didn't strip double quotes for you.

	HRB

(In my considered opinion) the DCL behaviour you describe is correct and
reasonable.  The problem is in how you're embedding the value of a string
within quotes; you should double any quotes that are in the string.  Sadly,
there's no easy way to do that from DCL.

This is a perfect example of where the ampersand (&) should be used, e.g.,
{MAIL/SUBJ=&P2}.  However, this doesn't seem to work correctly -- although
DCL doesn't complain, the quotes in P2 are removed.  But this may solve a
few problems.
					- Gilbert

Why did you put all those ''s in?   "Modern" DCL should be pretty well void
of := and of '.

For instance...

	if "''p1'" .eqs. ...

should be

	if p1 .eqs. ...



The best trick is how to define a symbol if it doesn't exist, but not change
it if it does.  I use

	MAKE_THIS_SYMBOL_EXIST[0,0] = 0


/Bevin

Sounds like .3 trades "''p1'" for p1[0,0] = 0 -- an even trade, but
thanks for the hint.

Still no real solution?  -- note I'm using NMAIL, not MAIL -- if that
makes a difference.

Martin.

I don't see how &P1 helps the problem any.  I pulled out the documentation
on it, and it seems to be just a delayed 'P1'.   What's really needed here
is a way get the literal sequence of characters typed for a parameter, 
complete with surrounding quotes and embedded double quotes.

.0 is probably like me.. I learned "''foo'" early on and the substitution 
rules for DCL were so persnickity I've done it exactly that way ever since.
	
	HRB

	The double-quote behavior in .COM files is an utter crock.
#6	<_Jym_>\

Consider the example:

	$! Does P1 contain a quote character?
	$!
	$ if f$locate("""",p1) .ge. f$length(p1) then write sys$output "Yes"

re .1 (HRB) and .6 (_Jym_)

	If DCL didn't strip double quotes for you, you may have trouble
	expressing the above example.  Or (depending on what alternate
	proposal you might use), you could have trouble with the following:

	$ x = "Please bullet the items with a ""+"" mark"
	$ x = "concatenate these " + "strings"
	$ if a .eqs. "" then if b .eqs. "foo" then goto label
	$ if a .eqs. """ then if b .eqs. ""foo" then goto label

	Many utilities ignore this problem.  For example, it's possible
	to create a CMS comment that will appear as two separate CMS commands
	in a CMS history (CMS automatically provides line breaks for these).
	The output of SHOW LOGICAL doesn't double quotes -- it's possible
	to create a logical name that is displayed ambiguously.  When Cobol
	extracts definitions from the CDD, producing Cobol source text, it
	*does* re-double the quotes for string literals (you *have* to get it
	right if the output is to be used by some nit-picky computer).

re .5 (HRB)

	I almost expected &P1 to do the trick.  It's a little different
	than a delayed 'P1'.  In the case of the mail example, if P2 is
	{foo "bar"}, then {MAIL/SUBJ=&P2} causes no DCL complaints, while
	a wide range of alternatives can or do cause errors.

	After seeing that {MAIL/SUBJ=&P2} gave no complaints, I was pleased,
	and expected to see {foo "bar"} in the subject line of the message.
	Unfortunately, the subject was given as {foo bar} -- how or why it
	managed to realize that P2 was a single entity, but failed to retain
	the embedded quotes, I don't know.  However, it *does* solve this
	particular problem (no DCL errors), if you're willing to forgive a
	few missing quotes from the subject of the message.

	Yes, what's needed is a way to get embedded quotes doubled again.
	It's a relatively simple matter to code a little loop to do this.
	A more convenient way to do this (perhaps another f$edit option)
	is desired.

					- Gilbert

RE .4		how to define a symbol if it doesn't exist, but
                not change it if it does.


$if f$type("symbol") .eqs. "" then symbol = "whatever"

Bye
Kay R. Fisher

What if you are running V3?

tom_p

If it's a number - it's easy

symbol = 0'symbol'

!If it wasn't defined before - it is now.  
!It is was - it has the same value.
!I used this for debugging DCL

Debug = 0'Debug'

Bye
Kay R. Fisher

What's wrong with:

	$ symbol = "''symbol'"

(Yes, the result is always a string; if you are know you need an integer you
can always do something like what is suggested in .-1.)

Note that, in this particular case, "=" and ":=" are equivalent.  What you
should NOT do is:

	$ symbol := 'symbol'

Besides screwing around with case, spacing, and quotes, and trying to execute
any substrings of symbol that look like calls to lexicals, this can have unex-
pected results if symbol ends with a "-".
							-- Jerry

$ symbol = " "" ! What a comment "

then

$ symbol = "''symbol'"  becomes

	 = " " ! What a comment "

so the value of the symbol is not preserved.

/Bevin

.12 brings us full circle back to my original question (if anyone cares):

HOW DO I PRESERVE QUOTES IN .COM FILES?

What kind of hackers are you?  It can't be that difficult.  (:-)

Martin.

$! send a wildcarded set of files to a destination using nmail
$! Two mandatory parameters:
$!	P1	file to send (argument to f$search())
$!	P2	destination (don't forget to quote arpa addresses)
$!
$	if P1 .eqs. "" then read sys$command P1 /prompt="Wildcard filespec to send: "
$	if P1 .eqs. "" then exit
$	if P2 .eqs. "" then read sys$command P2 /prompt="Destination (don't forget quotes): "
$	if P2 .eqs. "" then exit
$ loop:	file = f$search(P1)
$	if file .eqs. "" then exit
$
$	P0 = file
$	return = "goto 10$"
$	goto dq
$	10$: file = P0
$
$	P0 = P2
$	return = "goto 20$"
$	goto dq
$	20$: P2 = P0
$
$	nmail/notime  "''file'"  "''P2'"  "File ''file'"
$	goto loop
$
$ dq:	i = 0
$ dq1:	i = i + f$locate("""",f$extract(i,f$length(P0)-i,P0))
$	if i .ge. f$length(P0) then 'return
$	P0 = f$extract(0,i+1,P0) + f$extract(i,f$length(P0)-i,P0)
$	i = i + 2
$	goto dq1