is this intentional?

[Patrick Waugh]

Because if hg runs as the www user, then the per-user file (
/home/*/.hgrc ) of the actual user doing the work (ie the human) is
not being used.  In other words, it's all about context, and this is
not at all clear in the docs.

So, it is too easy to put something in a context where it will not
function.  This should be crystal clear in the docs.

In other words, functionally it is as I have described despite the
fact that the actual implementation may be different.  As a user, I
don't care how something is implemented, only how it appears to
function.  Obviously, deeper understanding helps, but the point of
being a user is not to have to understand how something is
implemented.  When my girlfriend uses her blow dryer, she does not
want to have to understand the implementation of the electromagnetic
engine which runs it.  =)

This is especially true since having a www user .hgrc file and its
behavior is mentioned in the manual only very briefly.

>From my perspective then, all I see is that a per-user setting (in my
home directory) fails to function.  Were it more clearly documented
that this is not the proper "context" for this option, along with the
proper context, then probleme solved.

Patrick


On Sat, Oct 4, 2008 at 7:29 AM, Peter Hosey <boredzo at gmail.com> wrote:
> On Oct 04, 2008, at 04:57:50, Patrick Waugh wrote:
>>
>> For example, it is not at all apparent that some [web] options can
>> function in the global, or per-repos hgrc files, while be useless in the
>> per-user config.
>
> That's because this isn't true. All the configuration directives go into the
> same pot (in a specific order), so if you put [web] options in the .hgrc
> file in the web server user's home directory, they will work just fine.
>
> Remember, Mercurial is running as the same user as the web server. As such,
> it looks in that user's home directory. There is no reason for it to look in
> any other user's home directory.
>
> The only exception might be in the case of a /~user URL, but I don't know
> whether it handles that case. (There's also REMOTE_USER, but the results of
> some Googling suggest that the value of that variable isn't necessary a user
> account on the machine.)
>
>