Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:86435
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: error (pb1.pair.com: domain belski.net from 85.214.73.107 cause and error)
To: "'Eric Stenson'" <ericsten@microsoft.com>,
	"'Ferenc Kovacs'" <tyra3l@gmail.com>,
	<internals-win@lists.php.net>
Cc: <internals@lists.php.net>
References: <BN1PR03MB105669DA9F95CD1E0B2BCDCD0C20@BN1PR03MB105.namprd03.prod.outlook.com> <CAH-PCH5LjKeK5RbE7UGoz_ANhLP89fBNtbjSdBnsNYuiPV=q4Q@mail.gmail.com> <002701d093a7$4d3a9210$e7afb630$@belski.net> <BN1PR03MB105E87E509BD6B07486FF65D0CC0@BN1PR03MB105.namprd03.prod.outlook.com>
In-Reply-To: <BN1PR03MB105E87E509BD6B07486FF65D0CC0@BN1PR03MB105.namprd03.prod.outlook.com>
Date: Fri, 29 May 2015 00:17:11 +0200
Message-ID: <000a01d09994$0e35f0d0$2aa1d270$@belski.net>
MIME-Version: 1.0
Content-Type: text/plain;
	charset="utf-8"
Content-Transfer-Encoding: quoted-printable
Thread-Index: AQGWHwiSIHHeGINRHuU4zDqP8dqMxwFrPrWcAgpiOsACDRVv2J3au+nQ
Content-Language: en-us
Subject: RE: [INTERNALS-WIN] Re: [PHP-DEV] Q: What is the Config.w32 equivalent of PHP_ADD_MAKEFILE_FRAGMENT?
From: anatol.php@belski.net ("Anatol Belski")

Hi Eric,

> -----Original Message-----
> From: Eric Stenson [mailto:ericsten@microsoft.com]
> Sent: Tuesday, May 26, 2015 6:18 PM
> To: Anatol Belski; 'Ferenc Kovacs'; internals-win@lists.php.net
> Cc: internals@lists.php.net
> Subject: RE: [INTERNALS-WIN] Re: [PHP-DEV] Q: What is the Config.w32
> equivalent of PHP_ADD_MAKEFILE_FRAGMENT?
>=20
> -----Original Message-----
> > From: Anatol Belski [mailto:anatol.php@belski.net]
> [...]
> > Can you please share what exactly you need it to do? Like the =
command
> > sequence or alike, maybe there is already a way to do that.
>=20
> I need to run the $(MC) command, with the '-um' option, which will =
create ETW
> message macros & binary resources.  The output is a .h, a .rc and a =
couple of
> .BIN files.
>=20
> e.g.:
> $(MC) -um -e h -h ..\pecl\wincache -r ..\pecl\wincache =
wincache_etw.man
>=20
> I'm working around the lack of a fragment by running the $(MC) command =
by
> hand, an checking in the results to the pecl\wincache directory, and =
hand-
> hacking the template.rc to #include the resulting wincache_etw.rc.  It =
kinda
> works, but it would be better if the building of the ETW resources was =
automatic
> with make/nmake.
>=20
> Even better would be to put the rendered .h, .rc & .bin files in the =
same directory
> as the .obj's, so they could be cleaned up.  The problem there is no =
way to pass
> an include directory arg (-I<dir>) to the $(RC) command used for =
template.rc.
>=20
I was just checking, one possibility were to make use of the variable =
called MFO which is globally available. Here for example what happens =
when you call EXTENSION(...) in the config.w32 =
http://git.php.net/?p=3Dphp-src.git;a=3Dblob;f=3Dwin32/build/confutils.js=
;h=3D9a7b3fa5ac990899189d537f8295a98be3b7f89c;hb=3DHEAD#l1290 . So this =
gives possibility to implement custom targets adding them through the =
javascript API. However I'm not sure one can add those targets as =
dependency to the main ext dll target, and this is neglecting almost any =
effect of creating a custom build target. What one could theoretically =
do is reading the whole Makefile and extending the target of the =
interest, then writing the whole back. However tricky it is and might =
break other exts, so should be done with care.

A bit another story is in master, there the objects are put into the =
response file, so theoretically one could do the necessary operation =
already when configure runs and extend the response file. But that =
doesn't really help globally. However it's tricky as well. We can do the =
necessary implementations in master and maybe backport it, it were  an =
improvement, but it were still missing in the already done releases. I'm =
going to put this on my todo at least for master. However don't expect =
coming to it very soon. Probably the simplest implementation were the =
possibility to add fragments as whole, disregarding the JS =
functionality. Maybe also you could play with the MFO possibility I've =
mentioned, I could have overseen something.

Regards

Anatol