 |
chamenas
Wizard
Joined: 26 Mar 2008
Posts: 1547
|
 Posted: Tue Jun 29, 2010 8:36 pm
3.21 BUG w/ Stringlists
|
To replicate:
| Code: |
<class name="test_string" id="121">
<var name="string_test" type="StringList" id="122">
<value>val1|val2|val3|val4</value>
<json>["val1","val2","val3","val4"]</json>
</var>
<trigger priority="15770" id="123">
<pattern>This is a test of (val%d)</pattern>
<value>$val_l=%trim( %replace( %1, "does", ""))
#show %if(%ismember( $val_l, string_test), "TRUE", "FALSE")
#show "$val_l: "$val_l
#SWITCH (%match( %trigger, %concat( "This is a test of ", $val_l)) && %ismember( $val_l, string_test)) {#show "MATCH!"}
{#show "NO MATCH!"}
#show %if(%ismember( $val_l, string_test), "TRUE", "FALSE")</value>
</trigger>
</class>
|
Install the script above, type in:
#show "This is a test of val1"
Your output should be:
This is a test of val1
TRUE
$val_l: val1
MATCH!
TRUE
This is as intended.
Now, close the session. Re-open the session and type in:
#show "This is a test of val1"
This is a test of val1
FALSE
$val_l: val1
NO MATCH!
FALSE
Now open the setting's editor and go to the test_string class. Click on the variable string_test, then click on the trigger.
Type:
#show "This is a test of val1"
You should get
This is a test of val1
TRUE
$val_l: val1
MATCH!
TRUE |
|
|
|
|
GeneralStonewall
Magician
Joined: 02 Feb 2004
Posts: 364
Location: USA
|
Posted: Tue Jun 29, 2010 9:28 pm
|
Any way you can simplify this?
|
|
|
|
|
chamenas
Wizard
Joined: 26 Mar 2008
Posts: 1547
|
Posted: Tue Jun 29, 2010 9:30 pm
|
Not sure, that was about as simple as I could make it when I looked at it. I could look again.
|
|
|
|
|
Tech
GURU
Joined: 18 Oct 2000
Posts: 2733
Location: Atlanta, USA
|
Posted: Wed Jun 30, 2010 1:29 pm
|
Confirmed (with caveats). So the behavior you described is very weird and certainly occurs with the code you posted.
Now the caveats. You are using %ismember but incorrectly. The syntax is as follows:
For example
| Code: |
| #IF (%ismember(@petitioner,@friendsList)) {cast heal @petitioner} {say Nope.} |
You are referencing the list name without the '@'. Now I may have missed a beta feature that allows this (or it may be a fluke of referencing session cache). Since this 'cache' presumably gets flushed when you close the session and the code fails. When you access the package editor, I suspect you repopulate this 'cache' and then the code works again. I'm sure Zugg can fill in the specifics but I'm pretty this is the gist of what's happening.
The test code works consistently with the '@' symbol.
| Code: |
<?xml version="1.0" encoding="ISO-8859-1" ?>
<cmud>
<class name="test_string" copy="yes">
<var name="string_test" type="StringList" copy="yes">
<value>val1|val2|val3|val4</value>
<json>["val1","val2","val3","val4"]</json>
</var>
<trigger priority="15770" copy="yes">
<pattern>This is a test of (val%d)</pattern>
<value>$val_l=%trim( %replace( %1, "does", ""))
#show %if(%ismember( $val_l, @string_test), "TRUE", "FALSE")
#show "$val_l: "$val_l
#SWITCH (%match( %trigger, %concat( "This is a test of ", $val_l)) && %ismember( $val_l, @string_test)) {#show "MATCH!"}
{#show "NO MATCH!"}
#show %if(%ismember( $val_l, @string_test), "TRUE", "FALSE")</value>
</trigger>
</class>
</cmud> |
|
|
_________________
Asati di tempari! |
 |
|
|
chamenas
Wizard
Joined: 26 Mar 2008
Posts: 1547
|
Posted: Wed Jun 30, 2010 3:11 pm
|
I was told dropping the @ symbol allows you to access the variable by reference, instead of making several copies of it, which would result in faster script execution for longer lists. However, if it's not saving the data, this is a problem.
It's likely that the same thing is happening here as well.
EDIT: Link to by-reference stuff. Am I misinterpreting it? |
|
|
|
|
Zugg
MASTER
Joined: 25 Sep 2000
Posts: 23379
Location: Colorado, USA
|
Posted: Wed Jun 30, 2010 4:24 pm
|
Chamenas is correct in his use of the "by reference" stuff. In 3.2x if you enter a function name into the command line and press F1 and see an argument of type "List", "ListRef", or "Table" then the argument allows the @ to be dropped for a direct reference that is faster than using @ (which makes a copy of the variable).
Still looking into the original bug report. |
|
|
|
|
Tech
GURU
Joined: 18 Oct 2000
Posts: 2733
Location: Atlanta, USA
|
Posted: Wed Jun 30, 2010 4:47 pm
|
| Tech wrote: |
| Now I may have missed a beta feature that allows this ... |
Man my memory is getting bad, because I was here for that other discussion too. 
Well Zugg's got the info he needs so it should get handled. |
|
_________________
Asati di tempari! |
|
|
|
Zugg
MASTER
Joined: 25 Sep 2000
Posts: 23379
Location: Colorado, USA
|
Posted: Thu Jul 01, 2010 12:23 am
|
There was a bug dealing with list/table variables not getting set properly at session startup. The internal json table was not being created until needed so sometimes a script could access this undefined table before it was created. Fixed for the next update.
|
|
|
|
|
|
|
|
