Mob_db2

Originally posted by [b]Kenpachi[/b]
What exactly is the bug?
Why should the server tell you that you entered to much data? The server doesn't care about this.
All the data which is required is read and everything else is left untouched.

//EDIT:
xazax just mentioned that it would be a nice feature to inform the user about extra columns.
-> Moved to Core section.

Originally posted by [b]Epoque[/b]
Processing columns is handled within sv_readdb() which receives a maximum number of columns to read from the database. As a result, sv_split() only processes however columns were specified as passed through sv_readdb(). I don't think there'd be a safe or reasonable method of splitting the columns without jeopardising some other components of the system.

[code]sv_readdb(db_path, filename[fi], ',', 31+2*MAX_MVP_DROP+2*MAX_MOB_DROP, 38+2*MAX_MVP_DROP+2*MAX_MOB_DROP, -1, &mob_readdb_sub);[/code]

As seen above, 38+2*MAX_MVP_DROP+2*MAX_MOB_DROP is the value used for maximum number of columns, sv_split() will only ever process this number of columns [i]unless [/i]we remove the safety check which might cause issues with other areas of rAthena (in terms of processing comma-seperated lines.)

Originally posted by [b]xazax[/b]
Strange..

sv_readdb already have a check, still it did not warn. Maybe max column number that is given is invalid?

[code]
if( columns > maxcols )
{
ShowError("sv_readdb: Too many columns in line %d of \"%s\" (found %d, maximum is %d).\n", lines, path, columns, maxcols );
continue; // too many columns
}
[/code]

Anyways itemdb does not use sv_readdb, and it does not check for excess comas.

Originally posted by [b]Epoque[/b]
Yeah, it could certainly be implemented on the item database, but as far as monsters go the sv_split() limits the total number of columns by decrementing the maximum number of fields per iteration. It stops processing as soon as it hits the maximum number of columns, and as such it'll never reach the code you pasted above (unless there's some serious, serious problem.)

Originally posted by [b]xazax[/b]
I just tested it. If I adjust the max size to the same as the min size, it will skip entries with too many columns, so sv_readdb does work well, just the function call is "broken". Extra columns are not used anyways, so I think it would be ok to lower the max column number to the same amount as min column number. Opinions?

Originally posted by [b]Gepard[/b]
This is a typo from [rev=15531]. SVN Blame is your friend.

Originally posted by [b]xazax[/b]
Fixed in [rev=15784].