No pretty screenshots for you today, just walls of text and some confused ramblings. I’m a little out of sorts and not really in any mental shape to be accomplishing interesting things. <\/p>\n
I’m feeling a bit un-creative and run down today. So I want to get away from the stuff that requires a lot of creativity (gameplay, writing, art) and just work on something straightforward and mechanical. I’m trying to make sure I do something on the project every day, but it’s probably a bad idea to do anything tricky while I’m all muddle-brained. I check out my miles-long to-do list and find something really dull: reading ini files.<\/p>\n
You know ini files. Stuff like this:<\/p>\n
<\/p>\n
\r\n[Settings]\r\nSetup=1\r\nmusic=0.40\r\nsound=1.00\r\nfullscreen=0\r\n[Window]\r\nWidth=1456\r\nHeight=887\r\n<\/pre>\nThis is important because right now my ini reader is the only part of my program that’s still tied to Windows. To read and write ini files I’m calling GetPrivateProfileString ()<\/code> and WritePrivateProfileString ()<\/code>. Boo. If I ever want to do a Linux port then I need to replace those with my own code. <\/p>\nYou might remember that last Christmas I wrote an ini file parser<\/a>. The good news is that the parser seemed to turn out okay. The bad news is that I was using Qt at the time. Which means I was using the much more useful and powerful Qt string class and not the bog-standard std::string. <\/p>\nCrud. <\/p>\n
std::string is how you’re supposed to use strings in C++. The problem is that it sucks. It doesn’t have a built-in way to strip whitespace off the beginning and end of a string, even though I have never ever written a parser of any sort that didn’t need that. Also, being able to compare two strings in a case-insensitive way is pretty fundamental to most text-processing problems. I need to find the [settings] section of the file even if it’s called [Settings] or [SETTINGS]. Also, I need to convert between numerical values and strings. Qt strings do all of these. std::string does none of them. <\/p>\n
So what happens? Everyone needs this stuff, so everyone writes their own string-manipulating tools to cover all these deficiencies. Then everyone has their own little set of string tools, which end up in all the code they write. So now if I want to share my ini reading code with somebody they also need my string tools. They likely have the exact same string tools, and now they have duplicates hanging off my project, cluttering up my code, duplicating their code, causing name collisions, and just generally making life less easy.<\/p>\n
I moaned about this on Twitter yesterday and got this reply:<\/p>\n
<\/p>\n\n@shamusyoung<\/a> "case" and lexical sorting are very complicated in I18N. C++ stdlib will never do it well because it can't marry unicode.<\/p>\n— James Iry (@jamesiry) October 8, 2013<\/a><\/p><\/blockquote>\n