[WildCard65]

Originally Posted by Powerlord Be aware that functag and funcenum don't exist in the new syntax... you need to look up what replaced them and use that instead. Edit: typedef replaced functag and union replaced funcenum. Speaking of which, I was going to work on passing a data arg to menus in the 1.7 core... I wonder if the AM team moved the menu code yet so I can start modifying them.

this? https://github.com/alliedmodders/sourcemod/pull/151

[friagram]

I mentioned this in the other post.

I've only had to do this once... basically what I did was use adt arrays, each cell having a handle to the forward.. function search name, an cipher (string), and then another arg as a datapack. It wasn't pretty but it got the job done. This was mainly because I wouldn't know what the thing would be doing until later, and if the other plugin would have the think functions loaded for that module (there were different think modules for different plugins). I guess you could use a string in stead or an any, and do away with the cipher, but you'll need to know what your forward is expecting, and it gets very messy very quickly.

Was like this

PHP Code:

new pack = CreateDataPack();
WritePackCell(entity);
WritePackCell(target);
WritePackFloat(velocity);

Function_Think("Surface to air homing", pack);

Function_Think(const String:method[], Handle:data)
{
  new index = FindStringInArray(gh_methods, method);
  if(method != -1)
  {
    decl String:cipher[64];
    GetArrayString(gh_cipher, index, cipher, 64);
    Call_StartForward(GetArrayCell(gh_forward, index))
    
    ResetPack(pack);
    for(new i; cipher[i] != '\0'; i++)
    {
       switch(cipher[i])
       {
            case 'i': Call_PushCell(ReadPackCell(pack));
            case 'f': Call_PushFloat(ReadPackFloat(pack));
            case 's':
            {
                decl String:temp[256];
                ReadPackString(pack, temp, 256);
                Call_PushString(temp);
            }
       }
    }
    Call_Finish();
  }
  else
  {
     // method wasn't found, do something...
  }
  CloseHandle(pack);

} 


As you can see, it's not pretty. And there are some assumptions:
You always know what you are going to do. If you call the function with the wrong cipher, you're screwed. I got around this by using the preproccessor and putting them in an include file, so they would always be the same.. but you still have to be sure that you that you pass all of the parameters or the forward will not get the data. This is okay if you don't know what modules you will be loading/unloading, and you can have the same "function name" do different things in different situations. However, if you start having multiple methods with the same name, you'll have to do something.. Perhaps store the plugin's handle AND library name that registered the forward if it's a global forward, and when that plugin's library is unloaded, remove it's data from your arrays. like deal with which one was loaded first, or figure out some kind of calling convention. This is a hacky solution, and one that I would avoid unless you really really need it.

[Dr. Greg House]

Take a look at sourceirc.

[Powerlord]

Originally Posted by WildCard65 this? https://github.com/alliedmodders/sourcemod/pull/151

Looks like it. I didn't think to check the pull requests to see if it was finished. Granted, I had just assumed it was when I merged master into my fork's master (which I subsequently merged into the branch I use for this).

[BAILOPAN]

You don't need any directives to use the new syntax - just the 1.7 compiler. You can still use functag/funcenum though we have introduced a new syntax and deprecated the old one:

typedef AbilityFunction function void();

The old syntax isn't "hard deprecated"... we just don't use it upstream anymore. It'll probably get deprecated in 1.8 and removed in 1.9, or something.

[nergal]

just an expansion on this thread

I tried to do this

PHP Code:

Function func = pack.ReadFunction();

int param1 = pack.ReadCell();
if ( param1 > 0 ) func(param1); 


and I get an error saying "invalid function call, not a valid address."

Is there seriously no other way to call a function without using call_startfunction?

[Powerlord]

Originally Posted by nergal just an expansion on this thread I tried to do this PHP Code: Function func = pack.ReadFunction(); int param1 = pack.ReadCell(); if ( param1 > 0 ) func(param1);  and I get an error saying "invalid function call, not a valid address." Is there seriously no other way to call a function without using call_startfunction?

iirc Function variables aren't closures. You have to use Call_StartFunction or put them into a Forward to call them.

[nergal]

Originally Posted by Powerlord iirc Function variables aren't closures. You have to use Call_StartFunction or put them into a Forward to call them.

would be interesting if in the future, there'll be a function class under Handles that can operate that way

[asherkin]

Originally Posted by nergal would be interesting if in the future, there'll be a function class under Handles that can operate that way

Some of the breaking changes made in the transitional syntax are for supporting that in the future (and suprise, suprise, people complain about the changes).

[nergal]

Originally Posted by asherkin Some of the breaking changes made in the transitional syntax are for supporting that in the future (and suprise, suprise, people complain about the changes).

If I can drag and shoehorn Ethics into this, wouldn't anything that makes it easier and enables further flexibility with sourcepawn better and worth it more than the people who complain about the changes?

Currently, I love the changes. Methodmaps do make it easier to code and it's alot cleaner than before. I don't have to write 'new' every damn time I want to make a simple variable.

Also, I'd really love to know who's complaining because all of these changes are for the better and make Pawn alot more flexible and expressive than before.

My only request in sourcepawn is having structs/classes and pointers but I understand that it'll be some time before they're introduced but they're worth the wait.

I'm assuming that there won't be classes/pointers until the old syntax is removed in order to maintain sourcepawn's speed?

Seriously, I would like to have a long chat with whoever complained and I'd love to know their complaints.