[Closed] Is there a getINISetting Key count limitation

Hmm – never heard of such a limit – but is it possible that you hit some general ini file limits ?
Don’t know wether this applies to 3ds Max/Maxscript ini files too, but i stumbled over this doc

INI files are ideal when the there is relatively little data (INI files have a 64K file size limit), and when it is in text-only format. INI files can be readily shared across a network simply by placing the file on a network drive. The main disadvantage to INI files is that the text for each key entry must be under 256 bytes (the example above limit the entry to 128 characters). Another disadvantage is that the data in an INI file can contain text only, no special formatting, tabs, hard returns, or extended formatting. INI files can be edited by any text editor or word processor.

http://www.gmccomb.com/vault/edge/ini.html

this is the sdk function that handles it calling the windows fn GetPrivateProfileString

Value*
 get_INI_setting_cf(Value** arg_list, int count)
 {
 	TCHAR val[INI_BUFFER_LEN];
 	if (count == 1 || count == 2) // get sections or section keys
 	{
 		TSTR  fileName = arg_list[0]->to_filename();
 		AssetId assetId;
 		if (IAssetManager::GetInstance()->StringToAssetId(fileName, assetId))
 			fileName = IAssetManager::GetInstance()->GetAsset(assetId).GetFileName();
 
 		GetPrivateProfileString(
 			(count == 2) ? arg_list[1]->to_string() : NULL, 
 			NULL, 
 			_T(""),	val, INI_BUFFER_LEN, 
 			fileName);
 		one_typed_value_local(Array* result); 
 		TCHAR *cp = val;
 		TCHAR tmpSec[INI_BUFFER_LEN] = {0};
 		int i;
 		vl.result = new Array(0);
 		while(*cp) {
 			i = 0;
 			while((tmpSec[i++] = *cp++) != _T('\0'));
 			vl.result->append(new String(tmpSec));
 			}
 		return_value(vl.result);
 	}
 	else
 	{
 		check_arg_count(getIniSetting, 3, count);
 		TSTR  fileName = arg_list[0]->to_filename();
 		AssetId assetId;
 		if (IAssetManager::GetInstance()->StringToAssetId(fileName, assetId))
 			fileName = IAssetManager::GetInstance()->GetAsset(assetId).GetFileName();
 
 		GetPrivateProfileString(
 			arg_list[1]->to_string(), 
 			arg_list[2]->to_string(), 
 			_T(""),	val, INI_BUFFER_LEN, 
 			fileName);
 		return new String(val);	
 	}
 }

have to not use the catchall then.

I didn’t realize that if you just passed a the file name to getinisetting it would just return the section headers. Makes for a good quick and dirty string table loader.

Hah, great input people!
So if that memory limit is for a single entry, and the mem limit for the entire returned array is huge, then there shouldn’t be a reason for the issue I’m having?
Very strange.

The problem is, that the script won’t know which entries exist in the INI till it gets the catchall array. The parameters stored may vary on a scene by scene basis.

I’ll try to break the data down into seperate “sections” and then parse through those before gathering the keys… assuming there isn’t a limit on the number of “sections” too!.

Yup, thanks again Lo.

Here’s a script to demonstrate the issue. :

no = 1000
inipath = "c:\	est.ini"
for i = 1 to no do (
	keystr = "keyval" + (i as string)
	setINISetting inipath "Sec" keystr "true"
)
testArr = #()
for i = 1 to no do (	
	keystr = "keyval" + (i as string)
	iniVal =  (getinisetting inipath "Sec" keystr )
	--format "% : %
" i (getinisetting inipath "Sec" keystr )
	append testArr iniVal
)

format "Specified entries : %
" testArr.count

testArr = (getinisetting inipath "Sec"  )
format "Catchall entries : %
" testArr.count

for i in testArr do format "%
" i

yeah the single data read is indeed massive though I wouldn’t really want to be polling an inifile hundreds of times, even if it’s very quick it’s just not the done thing

fn getSectionKeys filename sectionName =
(
	local allLines = (dotNetClass "System.IO.File").ReadAllLines filename
	local targetString = "[" + sectionName + "]"
	
	local headerIndex = 0
	for i = 1 to allLines.count while headerIndex == 0 do if allLines[i] == targetString do headerIndex = i
	local result = #()
	for i = headerIndex + 1 to allLines.count do
	(
		if allLines[i][1] == "[" do return result
		append result (trimRight ((filterString allLines[i] "=")[1]))
	)
	result
)

Why not just go with the XML solution? Sounds like it will be less head ache and work?