-
-
Notifications
You must be signed in to change notification settings - Fork 422
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add MAX_CONSOLE_LENGTH and COMMAND_MAX_* defines to console.inc #1732
base: master
Are you sure you want to change the base?
Conversation
This creates a define to match the underlying SourceMod buffer sizes for Print functions
I suggest adding defines by enum
{
COMMAND_MAX_ARGC = 64,
COMMAND_MAX_LENGTH = 512,
}; from CCommand |
Why is that? |
Сommand will not executed over the limit int nLen = Q_strlen( pCommand );
if ( nLen >= COMMAND_MAX_LENGTH - 1 )
{
Warning( "CCommand::Tokenize: Encountered command which overflows the tokenizer buffer.. Skipping!\n" );
return false;
} if( m_nArgc >= COMMAND_MAX_ARGC )
{
Warning( "CCommand::Tokenize: Encountered command which overflows the argument buffer.. Clamped!\n" );
} |
I see they are part of convar.h, are the values the same for all games? |
How do you envision these defines being used by plugins? Are there many plugins out there defining these values themselves? I believe the 1024 byte formatting buffer is completely arbitrarily sized on SM's side, and is meant to just be "bigger than any plugin could reasonably want to print on a single line" - there is no relation between these functions. |
My case bool ExecuteCommands()
{
bool bResult = this.hExecuteCommands != null;
if(bResult)
{
int iLength = this.hExecuteCommands.Length;
char sCommand[COMMAND_MAX_LENGTH]; // instead of the past #define MAX_COMMAND_LENGTH 512
for(int i = 0; i != iLength; i++)
{
this.hExecuteCommands.GetString(i, sCommand, sizeof(sCommand));
#if defined DEBUG
LogMessage("\"%s\"", sCommand);
#endif
InsertServerCommand("%s", sCommand);
}
if(iLength)
{
ServerExecute();
}
}
return bResult;
} |
@asherkin, I am of the opinion that if 512 bytes are not enough for the plugin, then let it be written not completely for the further command segment and output error from the handler, than will be incomprehensible
|
My use case in MAX_CONSOLE_LENGTH is for printing out large text chunks to the client/server console. I had to dig through SM code to figure out the max it allowed, which I strongly felt should have been defined somewhere. |
My only real gripe with this is that if we are going to be defining a maximum console length, we should have this be reflected in sm core as well. I don't like the idea of having constants exist in core that must remain identical to an SP counterpart without any kind of reference. |
@asherkin I think having some limit hidden in sm core with no reference for plugins isn't great, however arbitrary the actual size may be. Any information that requires a dive into sm core to figure out is probably something we should clarify for plugins. |
All the formatting that core does should probably be changed (where there isn't an actual practical and obvious limit) to use the fixed-size buffer initially, but if the limit is exceeded allocate a correctly sized buffer and use that for a 2nd attempt. It's either that or we just document it per-function, but unifying these together does not seem like the right approach. |
Here's a rough example I'm thinking of, probably along the lines of what you're suggesting? std::unique_ptr<char[]> dynamicFormat(void *pContext, const int *params, unsigned int param) {
size_t try_size = 1024;
auto str = std::make_unique<char[]>(try_size);
size_t len; // bytes written not including null term
while (true) {
DetectExceptions eh(pContext);
len = g_pSM->FormatString(str.get(), try_size - 2, pContext, params, param);
if (eh.HasException())
return 0;
if (len < try_size-1) {
break;
}
try_size *= 1.5;
str = std::make_unique<char[]>(try_size);
}
return str;
} Not sure how we'd know what the size would be by the second try |
The first try should be on a stack buffer to avoid the heap allocation by default. |
indeed. allocations like this will murder throughput. @Headline are you OK to take this on? |
This creates a define to match the underlying SourceMod buffer sizes for console related methods.
sourcemod/core/logic/smn_console.cpp
Line 91 in 237db05
sourcemod/core/logic/smn_console.cpp
Line 130 in 237db05
sourcemod/core/logic/smn_console.cpp
Line 155 in 237db05
sourcemod/core/logic/smn_console.cpp
Line 177 in 237db05
sourcemod/core/logic/smn_console.cpp
Line 266 in 237db05