SysInfo refactoring, startup log update

[Karlson2k]

Started as small startup log refactoring, this branch growed up to SysInfo refactoring and cleanup.
I fixed a lot of mess and inconsistency in SysInfo, removed some outdated and platform-specific functions.
New function are designed to be used in platform-agnostic way.

[Karlson2k]

jenkins build this please

[Karlson2k]

jenkins build this please

[Karlson2k]

@jmarshallnz Would you like to review? :)

[jmarshallnz]

Looks like you need to fixup atv/ios

[Karlson2k]

@jmarshallnz Thanks, that was stupid c/p error. :)
Fixed.
BTW darwin builds again missing NDEBUG define. Seems this includes Gotham release.

jenkins build this please

[Memphiz]

darwin builds have ndebug in release mode only - see

https://github.com/xbmc/xbmc/blob/master/tools/darwin/Configurations/Release.xcconfig

[Karlson2k]

@Memphiz xbmclogs full of logs with "Using Unknown XBMC x32 build" on darwin.
Do we have other choice than Release and Debug?

[MilhouseVH]

Seems to work fine on Raspberry Pi (OpenELEC).

Before PR4811 log
After PR4811 log

The only noticeable change on the System Info page (Confluence) is the change of operating system name from:
Linux 3.14.4 #1 PREEMPT Thu May 29 02:25:30 BST 2014 armv6l GNU/Linux
to
OpenELEC_Helix 0.0 (kernel: Linux ARM 32-bit version 3.14.4)
Note sure what the "0.0" following OpenELEC_Helix is about though.

[stefansaraev]

@MilhouseVH probably something we miss in os-release. thanks for pointing it. I'll take a look.

[Karlson2k]

@stefansaraev @MilhouseVH Right, you need it in /etc/os-release

[MilhouseVH]

Just a guess, but it probably needs to be the OE git hash or something that uniquely identifies the OE build as neither the log or Sys Info screen identify the OE build beyond "OpenELEC_Helix" where as before the log at least included the full OE name (build date, time, revision, git hash).

[stefansaraev]

@MilhouseVH I'd like to move the non-related openelec discussion in irc, to avoid spaming others. I'll check whats required and do a PR for OE, or ping you in irc.

[stefansaraev]

ok @Karlson2k I believe there is something wrong in os-release parser, getValueFromOs_release.
in fact, it doesnt work even with ubuntu's shipped os-release (14.04). and it even says "Unknown Linux Distribution", if I move VERSION_ID= on TOP of the file (first line, so NAME= is second). hum.

[Karlson2k]

@stefansaraev Thanks, I'll check it

[Karlson2k]

@stefansaraev Did you disable lsb-release parser getValueFromLsb_release() for your tests?
It must work as fallback and return correct values when /etc/os-release parser fails. lsb-release parser works fine on my Ubuntu.

[stefansaraev]

nope. I did not. our lsb_release returns nonsense anyway ;) it doesnt support -v/-c/etc.
tbh I didnt expect the parser to fail reading perfectly valid os-release file

EDIT: hint for a simple os-release parser: https://lkml.org/lkml/2012/9/5/419 maybe this helps.

EDIT2: to be clear, I did not test on ubuntu. I tested on openelec with /etc/os-release c/p'd from ubuntu (I thought there is something wrong with OE's os-release, which is fine, except not wrapping the values with "")

[Karlson2k]

@stefansaraev Thanks, I fixed this bug.
Should works well now.

jenkins build this please

[Karlson2k]

@stefansaraev According to http://www.freedesktop.org/software/systemd/man/os-release.html you must quote values if they contain spaces.
os-release must be compatible with bash syntax as it may be included in scripts like

#!/bin/bash
. /etc/os-release

[MilhouseVH]

Building OpenELEC with the latest commits and I'm getting this build error:

Application.cpp: In member function 'virtual bool CApplication::Create()':
Application.cpp:719:118: error: request for member 'c_str' in 'g_sysinfo.CSysInfo::GetUsedCompilerNameAndVer', which is of non-class type 'std::string() {aka std::basic_string<char>()}'
   CLog::Log(LOGNOTICE, "XBMC compiled " __DATE__ " by %s for %s %s %d-bit %s (%s)", g_sysinfo.GetUsedCompilerNameAndVer.c_str(), g_sysinfo.GetBuildTargetPlatformName().c_str(),
                                                                                                                      ^
/home/neil/projects/OpenELEC.tv/build.OpenELEC_Helix-RPi.arm-devel/xbmc-master-14-5ec51aa/Makefile.include:93: recipe for target 'Application.o' failed
make[2]: *** [Application.o] Error 1
make[2]: *** Waiting for unfinished jobs....

[Karlson2k]

@t-nelson Lowered number of commits.

[Memphiz]

@Karlson2k seems i am blind - darwin sets _DEBUG (with underscore) of course. Seems impossible to me that "buildType = "Unknown";" could be triggered in Application.cpp with the code i am seeing in that PR here. wonders

[Karlson2k]

@Memphiz See xbmclogs.com - a lot of Gotham logs wtih "Unknown" type.

[Memphiz]

@Karlson2k thx for the headsup - found the problem - pr incoming...

[Karlson2k]

This is the reason why I always add third case when only two choices must be valid. :)

[Memphiz]

@Karlson2k for reference #4851

[MartijnKaijser]

good for merge (after jenkins build) or does it need some squashing?

[Karlson2k]

@MartijnKaijser It's squashed enough and ready for merge.

[Karlson2k]

jenkins build this please just in case

[Karlson2k]

Anyone want to review? @jmarshallnz ? :)

[jmarshallnz]

Looks OK to me on the whole. Didn't check all the hundreds of little changes though for sanity/null check/running off array. I'm sure you'll fix them if there are any that snuck in.