Re: [lmi] [lmi-commits] master ba65912 5/6: Include the appropriate PCH header

[Greg Chicares]

On 2017-03-14 23:59, Vadim Zeitlin wrote:
> On Mon, 13 Mar 2017 09:46:14 -0400 (EDT) Greg Chicares <address@hidden> wrote:
> 
> GC> branch: master
> GC> commit ba65912f5d16ce8f626388d3bddb8e9641a5f319
> GC> Author: Gregory W. Chicares <address@hidden>
> GC> Commit: Gregory W. Chicares <address@hidden>
> GC> 
> GC>     Include the appropriate PCH header
> GC>     
> GC>     Formerly, the wx PCH header was included, but this file does not 
> depend
> GC>     on wx. Now it includes the non-wx PCH header instead.
> GC> ---
> GC>  msw_workarounds.cpp | 2 +-
> GC>  1 file changed, 1 insertion(+), 1 deletion(-)
[...]
>  Sorry, but I've belatedly realized that this broke MSVC build because, as
> I just wrote in my commit message:
> 
>     Revert the PCH header change
> 
>     This reverts ba65912f5d16ce8f626388d3bddb8e9641a5f319: while this file
>     doesn't use wx on its own, it is still included in libskeleton, in which
>     all files use pchfile_wx.hpp as the precompiled header and not doing
>     this breaks MSVC build.
> 
> I could work around this by disabling PCH use for this particular file, but
> it would be simpler to just revert this commit, using a simple "git revert"
> with your own commit message or by applying the attached (git-am-compliant)
> patch.

Reverted. To be pushed ere long. BTW, there was no attachment.

> This would also be consistent with the commit message of
> c2d02f6f4a6fdf33b2329c7140a4dd06904de281: "Use pchfile_wx.hpp in the files
> which are part of libskeleton" and this rule about which files should use
> pchfile_wx.hpp and which should not seems simpler to me than the one about
> trying to understand whether the file does or doesn't (even indirectly!)
> uses any wxWidgets functionality.

I've also updated the PCH headers to state this rule in both of them.
Before committing ba65912, I consulted those headers, but not the git
log history.