[Vlisp] What to do with 'command-s'

[Simon_Weel]

There's a post about the behavior of AutoCAD 2015 and (command ...) functions in custom *error* functions. AutoCAD 2015 displays this message if it encounters this situation: "Cannot invoke (command) from *error* without prior call to (*push-error-using-command*). Converting (command) calls to (command-s) is recommended." So I edited all LISP files and changed (command ...) to (command-s ...). And then I found out this doesn't work in AutoCAD 2010. Command-s was introduced in release 2012. We use different versions of AutoCAD and they all use the same LISP routines. I would like to keep it that way, so I wonder if I can either get rid of the 2015-message or tweak things to get (command-s ...) working in release 2010. Any ideas? Simon

[hmsilva]

---

@Simon_Weel wrote:
There's a post about the behavior of AutoCAD 2015 and (command ...) functions in custom *error* functions. AutoCAD 2015 displays this message if it encounters this situation: "Cannot invoke (command) from *error* without prior call to (*push-error-using-command*). Converting (command) calls to (command-s) is recommended." So I edited all LISP files and changed (command ...) to (command-s ...). And then I found out this doesn't work in AutoCAD 2010. Command-s was introduced in release 2012. We use different versions of AutoCAD and they all use the same LISP routines. I would like to keep it that way, so I wonder if I can either get rid of the 2015-message or tweak things to get (command-s ...) working in release 2010. Any ideas? Simon

---

Just a thought, untest (I don't have 2015 in this laptop) try to change the command function to vl-cmdf function (vl-load-com first)...

HTH

Henrique

[Gary_J_Orr]

try performing a version check prior to calling the command/command-s so you can call the appropriate version...


(if (>= (atof (getvar "ACADVER")) 18.2)
(command-s ...)
(command ...)
)

-Gary

[Simon_Weel]

I tried 'vl-cmdf'. Replaced (command ...) in all LISP routines. But the result is the same, which is kind of weird, since the routines no longer use (command ...). Double checked the code to see if one had escaped me. Looks like vl-cmdf evaluates the code and then passes it on to (command ...)? Simon

[owenwengerd]

This works in all versions:

((if command-s command-s command) "XXXX" "XXXX")

[Gary_J_Orr]

one could... but you would also need to consider a call to (initcommandversion) to ensure that all things are equal when passing the command options to the command...
My suggestion allows for an easy method to build seperate command strings if need be to account for such... and if the same command strings can be sent to both, well copy paste can duplicate the strings in both versions while allowing for an easy method to modify either as things change even more going forward.

-G

[dbroad]

I like it Owen.  If a file used command-s more than once in a file, I wonder if it would be more convenient to put one of these at the top of the file:

;;backward compatibility for earlier versions

(if (not command-s) (setq command-s command))

or 

;;backward compatibility for earlier versions

(or command-s (setq command-s command))

or for the really paranoid to head of the case of a non-functional binding of command-s

;;backward compatibility for earlier versions

(if (/= 'subr (type command-s)) (setq command-s command))

Then the rest of the file could be written as normal.

For the benefit of the the casual readers of this thread, be reminded that command-s is only for complete command sequences.  No user input can be nested into the arguments to the command-s function.

[owenwengerd]

---

@dbroad3 wrote:

If a file used command-s more than once in a file, I wonder if it would be more convenient to put one of these at the top of the file:

Once could of course, but for code readability and debuggability, I prefer to see the logic at the point where it is used.

[dbroad]

Another good point.

[dgorsman]

Could also build it as a utility wrapper defun e.g. (defun command-i ( arg_list / )...), although the arguments would need to be (list ...) 'ed rather than a variable number of strings.  Opens up a few additional interesting thought, like global variable(s) with re-useable command argument lists for common command sequences.

[Gary_J_Orr]

oops, wrong version... command-s was introduced in 2013 (ver 19) it seems...
G

[hmsilva]

---

@Anonymous wrote:
oops, wrong version... command-s was introduced in 2013 (ver 19) it seems...
G

---

Garry,

command-s was introduced in 2012, but only referenced in 2013 Help files...

Henrique

[Gary_J_Orr]

thanks Henrique...
I had thought so but couldn't find proof so...
and it seems that it wasn't a "forced requirement" for command functions within the Error handler (or other post-processing functions) until more recently anyway (based upon my 2012 not having that particular warning when my error handlers kick in, either that or I've managed to catch all my errors in my code and it just hasn't impacted me)...
But it's good to know for future reference
-G

[dbroad]

I guess that is possible but I would probably not use such a function.  Over the years, I've built up my own way of handling defaults.  If you develop the thought further, I might be interested.

[Simon_Weel · Accepted solution]

No, it doesn't work in 2010. However ((if command-s command-s vl-cmdf) "XXXX" "XXXX") DOES work. So problem fixed! Simon

[JamesMaeding]

I know this is months past this thread, but

((if command-s command-s command) "XXXX" "XXXX")

does not work in acad 2009. It gives an odd error.

[sinisterkid]

Sounds like you guys all have really technical solutions, which I would appreciate if i had to hack further. I had this problem with the CLIPIT command (cropping some jpeg images) and I merely closed out of cad, regened, and reopened. Problem solved. just a thought before going to town on a LISP.

[JamesMaeding]

@sinisterkid

That kind of answer makes support people look bad though.

You might get past one problem, but people start to lose faith in you if you don't fix things right, so hence the search for a real fix here.

[molly.kumer2016]

Hi Simon, 

I understand this is an old thread, but I hope you'll see my message! I encountered a similar problem with a lisp recently. all of a sudden, when trying to use a lisp which worked just fine until now, I get the message you mention. I have AutoCAD 2015 and would like to try edit all LISP files and change (command ...) to (command-s ...). Can you please tell me how? 

desperateMolly

[roland.r71]

Just check each custom/local *error* function inside your LISPs, and change each (command "func" "arg") in (command-s "func" "arg"). (You can (SHOULD) leave the rest of the lisp alone !)

For the record, command-s became mandatory for *error* functions as of ACAD2015 (R20), so to keep it working in both newer and older versions, i use a version check myself (as has been mentioned before in this thread).

Example:

   (defun *error* ( msg / )
      (if (< (atoi (getvar "acadver")) 20) ; ACAD 2014 or older
         (command   "_.UNDO" "_End") ; End Undo - old version
      ; else
         (command-s "_.UNDO" "_End") ; End Undo - new version
      )
; rest of error function here...

edit:

...and don't mind @sinisterkid as he's comparing apples & oranges.

You can restart, regen and/or reopen until you weight about one ounce, but 2015 & higher will simply keep telling you it is not allowed to use (command ...) within a *error* function.

Just like 2011 & older will keep telling you it doesn't recognize the command (command-s ...)

(2012-2014 are 'hybrids'. They have the command-s, but they will (still) accept command for *error* functions.