[Outdated] PortableApps.com Launcher 2.0 Beta 4

John just helped me to pin down the crash issue Scriptdaemon was experiencing in certain obscure situations of the INI file in OpenTTD Portable on Windows 7. Replacing the new, kernel32::GetPrivateProfileSection ForEachINIPair.nsh with the old, FileRead etc. ForEachINIPair.nsh stopped it from crashing. So I've gone back to the old ForEachINIPair macro plus a couple of extras which I should have had earlier to make it behave as expected, whitespace trimming and quotation mark parsing; this means that a line like `foo = " x"` evaluated to `foo ` and ` " x"` with the old ForEachINIPair macro, but with the new GetPrivateProfileSection one and my update to the old one (i.e. the current one in the repository and what will be in the next release), it'll be `foo` and ` x` as expected.

This means that launcher.ini can no longer be encoded in UTF-16LE, but I ahve yet to see a case where that will be necessary, anyway. At some point in the future I'll get round to making it capable of dealing with either (I'll refer back to Gringoloco's patch) but for the moment it's ASCII-only.

Now I'm happy to release a Release Candidate (but I'll wait for Simeon's . registry issue if he's quick).

Scriptdaemon: the important thing in your report that language switching isn't working properly is that it's saying that the environment variable PortableApps.comLocaleName is empty (set, but empty). The problem is, I can't work out why that would be. Can you see anything wrong with this code?

	ReadEnvStr $0 PortableApps.comLocaleName
	${If} $0 == ""
		ReadEnvStr $0 PortableApps.comLocaleWinName
		StrCpy $0 $0 "" 5 ; Chop off the LANG_
		${SetEnvironmentVariable} PortableApps.comLocaleName $0
	${EndIf}

That's all the code that there is for PortableApps.comLocaleName. I presume PortableApps.comLocaleWinName is LANG_GREEK? (Refer to Process Explorer to inspect the environment, could be useful if you could tell me the values of all of them inside OpenTTD.)

For those that can't be bothered getting Mercurial, they can now try out the hg tip by downloading an archive in bzip2, gzip or zip format, from the development repository; see the "bz2 | zip | gz" links near the top.

No excuses now.

(Explanation: I finally got round to adding the [web] value allow_archive = gz zip bz2 to .hg/hgrc in the SourceForge repository.)

A new release (an RC) is due some time soon. I think the only remaining known issues are with the registry plugin (and not very serious issues at that, missing Wine 1.0 support and registry keys with a dot (.) in them not working). Any last issues people have which I've missed (I was not present over the weekend and quite a bit happened over the whole site during that time...), please report them now or they probably won't be fixed for RC1.

I finally tracked down the issue with the language variables. Relative jumps don't just skip over !defines which !insertmacros, so it was landing in the middle of the SetEnvironmentVariable macro call, and so its pushing for use by kernel32::SetEnvironmentVariable was not complete yet, and it lost one of its variables and was left with an empty variable (because I'd pushed $_LOGICLIB_TEMP which was empty though there was no need to).

I tested this thoroughly with OpenTTD (hey, don't those Greek characters look strange... [explanation to avoid misunderstanding: funny little characters, not incorrect characters]) and it's all working now.