lmi
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lmi] Replacing boost with C++ standard library


From: Vadim Zeitlin
Subject: Re: [lmi] Replacing boost with C++ standard library
Date: Tue, 27 Feb 2018 02:23:14 +0100

On Mon, 26 Feb 2018 23:02:55 +0000 Greg Chicares <address@hidden> wrote:

GC> On 2018-02-26 22:02, Vadim Zeitlin wrote:
GC> > On Mon, 26 Feb 2018 20:46:26 +0000 Greg Chicares <address@hidden> wrote:
GC> [...]
GC> > GC> >  I could also resurrect my ancient patch updating the code to use 
version 3
GC> > GC> > (on which C++17 std::filesystem is based) of Boost.Filesystem API 
instead
GC> > GC> > of the version 1 that we use currently, not so much for applying 
it, but to
GC> > GC> > use as a base for the later switch to std::filesystem.
GC> > GC> 
GC> > GC> Yes, please!
GC> > 
GC> >  I've looked at this and while the original patch still applies more or
GC> > less cleanly, it's not at all enough to build with std::fs, there are many
GC> > more changes needed on top of the ones I had already done. I'll work on
GC> > them, but I'm afraid the resulting patch will end up being quite difficult
GC> > to review just because of the sheer number of changes, even if most of 
them
GC> > are quite trivial.
GC> 
GC> Is it that there are only a handful of conceptual changes, but they
GC> affect code in many places in similar ways?

 There is potentially more than a handful of changes, and most of them
indeed need to be repeated in many places in the code.

GC> One idea that seems obvious is to keep using namespace 'fs' as an
GC> intermediate step.

 Oh, yes, sure, I didn't even think of changing this.

GC> That occurs on 342 lines. It'll be easy enough to change it globally to
GC> 'std' later.

 I am not sure we should change it, std::filesystem is a mouthful. And I
don't think it's legal to use std::fs as an alias.

GC> >  OTOH I don't think it makes sense to ask you to review a patch for
GC> > building with Boost.Filesystem v3 separately, if the goal is to switch
GC> > directly to std::filesystem anyhow, does it?
GC> 
GC> No, I don't imagine it makes sense to use a later boost version as
GC> a stepping-stone.

 Thanks for confirming this, I'll submit all the changes together at the
end then -- but it will need some more work and time.

VZ


reply via email to

[Prev in Thread] Current Thread [Next in Thread]