Re: [edk2-devel] [edk2-stable202311][PATCH] BaseTools: Python VfrCompiler implementation

["Michael D Kinney" <michael.d.kinney@intel.com>]

There are several advantages for this direction:


* Current VFR compiler in C has dependencies on very old libs that
  have not been updated.

* The movement to python will remove the pre-build step that requires
  some of the build tools to be built using host C compiler before
  running edk2 build command.

* The other element is moving all the python code into edk2-basetools
  repo with a published pip package.  This enables the use of 
  pip-requirements.txt to provide developers all the content needed
  to build.

I agree that we should not have both VFR compilers.  We need to make
sure the new one in Python is 100% compatible with the C version and
make the decision to simultaneously add Python one and delete the C
one and commit to the Python one.  I provided this feedback to the
VFR developers in the TianoCore Tools/CI meeting earlier this year.

The perf question is very good. It would be good for the VFR developers
to provide some perf comparisons.  I do not expect any significant
different that would impact overall platform build times.

Mike

> -----Original Message-----
> From: Pedro Falcato <pedro.falcato@gmail.com>
> Sent: Friday, December 15, 2023 9:04 AM
> To: devel@edk2.groups.io; Chen, Christine <yuwei.chen@intel.com>
> Cc: Gao, Liming <gaoliming@byosoft.com.cn>; Rebecca Cran
> <rebecca@bsdio.com>; Zimmer, Vincent <vincent.zimmer@intel.com>; Kinney,
> Michael D <michael.d.kinney@intel.com>; Leif Lindholm
> <quic_llindhol@quicinc.com>; Andrew Fish <afish@apple.com>; Feng, Bob C
> <bob.c.feng@intel.com>; Yang, Yuting2 <yuting2.yang@intel.com>; Hartung,
> Stephen <stephen.hartung@intel.com>
> Subject: Re: [edk2-devel] [edk2-stable202311][PATCH] BaseTools: Python
> VfrCompiler implementation
> 
> On Thu, Dec 7, 2023 at 9:08 AM Yuwei Chen <yuwei.chen@intel.com> wrote:
> >
> > Hi Liming,
> >
> >
> >
> > Is this feature been tested and reviewed these two weeks? 😊
> 
> Two questions:
> 
> 1) What testing strategy do you have to test for regressions in such a
> huge rewrite?
> 2) What's the point in shipping this to upstream if you're not aiming
> for the replacement of the original VfrCompiler?
> 3) What's the value of rewriting this in Python? If the existing
> VfrCompiler is already working fine (AFAIK?), a python version will
> likely just be slower (unless the original C version is super badly
> written).
> I *seriously* struggle to understand what this Python movement is
> supposed to do, except gratuitously rewrite large bits of BaseTools
> for a net loss (performance)
> 
> --
> Pedro


-=-=-=-=-=-=-=-=-=-=-=-
Groups.io Links: You receive all messages sent to this group.
View/Reply Online (#112605): https://edk2.groups.io/g/devel/message/112605
Mute This Topic: https://groups.io/mt/102486097/7686176
Group Owner: devel+owner@edk2.groups.io
Unsubscribe: https://edk2.groups.io/g/devel/leave/12367111/7686176/1913456212/xyzzy [rebecca@openfw.io]
-=-=-=-=-=-=-=-=-=-=-=-