Re: [lmi] [lmi-commits] (no subject)

[Greg Chicares]

On 2016-06-05 15:36, Vadim Zeitlin wrote:
> On Sun, 5 Jun 2016 03:17:05 +0000 (UTC) Greg Chicares <address@hidden> wrote:
> 
> GC> branch: master
> GC> commit 5bf279a08589d56e34f0afc6268a720939689cbb
> GC> Author: Gregory W. Chicares <address@hidden>
> GC> Date:   Sun Jun 5 01:08:45 2016 +0000
> GC> 
> GC>     Regularize borland conditionals
> GC>     
> GC>     This old compiler is not supported, but the conditional block
> GC>     at the beginning of every '.cpp' file may be useful in future
> GC>     for precompiled headers with other compilers
> 
>  Does this mean I could propose a patch enabling PCH for MSVC now?

Probably. Could you show me a simple conceptual proposal, e.g., a
diff for a single file? I was thinking of doing something like:

-#if defined __BORLANDC__
+#if defined LMI_USE_PCH
 #   include "pchfile.hpp"
 #   pragma hdrstop
-#endif // defined __BORLANDC__
+#endif // defined LMI_USE_PCH

i.e. 's/__BORLANDC__/LMI_USE_PCH/' for these PCH blocks only.

I think that's the gist of the PCH changes we've discussed in the
past. If I've remembered that correctly, then just tell me whether
'LMI_USE_PCH' is the right string to use, and I'll apply the change
myself (because I'm making sweeping changes anyway, and if I do
them all, there won't be collisions).

IIRC, the old PCH patch was something like that, along with big
changes to a very small set of files (of cardinality one?) that
indicate which headers to precompile. If I've remembered that
correctly, then that small part could be a separate patch that you
could provide at your leisure.

> This is
> the main (and practically only) remaining difference between my branch and
> master, but it's a pretty extensive one as it affects all .cpp files. This
> also means that I have 234 conflicts when trying to merge with the latest
> master, so if I could make this patch now, it would be really convenient as
> I have to update all the files anyhow.

Okay, sounds good--show me a diff of a simple file, and I'll make
the change globally. Using something other than '__BORLANDC__' to
conditionalize PCH inclusion will make it easier to identify the
uses of that macro that really do mark ancient workarounds, so that
they can be eliminated.

(BTW, I'm leaving some workarounds in place because they seem to
indicate genuine differences rather than standard-nonconformance.
Even though borland will probably never work, some other compiler
might need similar workarounds.)