Remove g_get_home_dir()

Remove g_get_home_dir()

$HOME is always set, and the xdg spec only wants $HOME, so remove a glib
call.

http://standards.freedesktop.org/basedir-spec/basedir-spec-latest.html

R=thestig@chromium.org

Committed: https://crrev.com/b36c7065b2b4ae075db74e94bd2b80fa45e253a5
Cr-Commit-Position: refs/heads/master@{#348819}

[rvargas (doing something else), 2015-09-14 22:21:40]

rvargas@chromium.org changed reviewers:
+ thestig@chromium.org
- rvargas@chromium.org

[rvargas (doing something else), 2015-09-14 22:24:54]

rvargas@chromium.org changed reviewers:
+ rvargas@chromium.org

[rvargas (doing something else), 2015-09-14 22:24:54]

Lei, do you mind reviewing this?

[rvargas (doing something else), 2015-09-14 22:25:49]

rvargas@chromium.org changed reviewers:
- rvargas@chromium.org

[Lei Zhang, 2015-09-14 22:43:22]

On 2015/09/14 22:24:54, rvargas (slow to respond) wrote:
> Lei, do you mind reviewing this?

Sure. I'm curious what's the motivation for this CL? Is it simply that if one
interprets the XDG spec as saying $HOME is always set, so this is effectively
dead code?

[Lei Zhang, 2015-09-14 22:44:39]

And for the record, g_get_home_dir() was added 6.6 years ago in r9162. Wee!

[knthzh, 2015-09-14 23:17:11]

On 2015/09/14 22:43:22, Lei Zhang wrote:
> Sure. I'm curious what's the motivation for this CL? Is it simply that if one
> interprets the XDG spec as saying $HOME is always set, so this is effectively
> dead code?
Yup.

> And for the record, g_get_home_dir() was added 6.6 years ago in r9162. Wee!
Ah, technology. How things change in just a few years!

[Lei Zhang, 2015-09-14 23:38:56]

My reading of the XDG spec is that it falls back to $HOME/foo when
$XDG_DATA_HOME and $XDG_CONFIG_HOME are not set. i.e. If $XDG_{CONFIG,DATA}_HOME
is set, XDG does not actually care about $HOME. Also, if $HOME is empty,
$HOME/foo just evals to /foo, which is still a valid, but likely not writable
path. So $HOME "should" be set, but does not "must" be set.

My take is that on one hand, it does seem silly to not have $HOME set, so that's
a vote for this CL.

OTOH, knowing Linux users, a few out there probably doesn't have it set, for
whatever reason. If it's just 6 lines of code is the difference between making
these silly users happy and breaking Chromium for them, maybe just leave this
alone?

I tried unsetting $HOME, and Chromium launches, prints a few errors, but seems
to work for the most part.

[knthzh, 2015-09-15 00:21:25]

There is also a stackexchange on this, although I didn't want to put that in a
commit message:
http://unix.stackexchange.com/questions/123858/is-the-home-environment-variab...

Chrome does falls back to /tmp, which would probably be closer to what a user
without $HOME wants.

Mainly I want to remove as many glib calls as possible, and maybe kill the glib
dependency someday (even if it is hidden behind a USE_GLIB macro, having
multiple build configurations for one platform is just silly).

[Lei Zhang, 2015-09-15 00:44:32]

On 2015/09/15 00:21:25, knthzh wrote:
> There is also a stackexchange on this, although I didn't want to put that in a
> commit message:
>
http://unix.stackexchange.com/questions/123858/is-the-home-environment-variab...
> 
> Chrome does falls back to /tmp, which would probably be closer to what a user
> without $HOME wants.
> 
> Mainly I want to remove as many glib calls as possible, and maybe kill the
glib
> dependency someday (even if it is hidden behind a USE_GLIB macro, having
> multiple build configurations for one platform is just silly).

I don't think stackexchange is the authoritative answer, but it does link to the
Open Group Base Specifications, which I find more convincing. Specifically:

http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap08.html

HOME: The system shall initialize this variable at the time of login to be a
pathname of the user's home directory. See <pwd.h>.

http://pubs.opengroup.org/onlinepubs/9699919799/basedefs/V1_chap01.html

shall: For an implementation that conforms to POSIX.1-2008, describes a feature
or behavior that is mandatory. An application can rely on the existence of the
feature or behavior.


Given the above, I'm going to say lgtm. When the users we break yell at us, we
can tell them to follow the POSIX standard. :)

[knthzh, 2015-09-15 00:51:25]

The CQ bit was checked by knthzh@gmail.com

[commit-bot: I haz the power, 2015-09-15 00:52:11]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1339573003/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1339573003/1

[commit-bot: I haz the power, 2015-09-15 02:09:46]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-09-15 02:10:34]

Patchset 1 (id:??) landed as
https://crrev.com/b36c7065b2b4ae075db74e94bd2b80fa45e253a5
Cr-Commit-Position: refs/heads/master@{#348819}

[commit-bot: I haz the power, 2015-09-23 12:41:35]

Patchset 1 (id:??) landed as
https://crrev.com/b36c7065b2b4ae075db74e94bd2b80fa45e253a5
Cr-Commit-Position: refs/heads/master@{#348819}