On 2022/11/01 13:57, Volker Schlecht wrote:
> Here's an updated difftastic port:
>
> * Update to 0.37.0
> This should among other things improve memory usage significantly
>
> * Add a MESSAGE concerning the possible need to raise datasize-max to
> reasonably use difftastic. I didn't benchmark it myself, but I don't think
> that the optimizations in 0.37.0 make it universally usable with a 4GB
> datasize
>
> * Include generation of html docs as proposed by Laurent Cheylus
>
> Anyone to review and import?
There are some whitespace nits in DOCDIR and the continuation line
in post-extract, otherwise OK sthen to import
I agree with devel as a category, this is not a text diffing program
> On 10/26/22 12:39, Stuart Henderson wrote:
> > This requires very large amounts of memory at runtime, it's crashing
> > for me with various input files with the 4GB datasize limit that is
> > the default for the 'staff' group (often leaving a just-under 4GB core).
> >
> > If there's nothing that can be done about the memory usage, is it
> > worth adding a warning in MESSAGE?
> >
> > It seems a lot happier with 8GB available.
> >
> > #0 thrkill () at /tmp/-:3
> > #1 0x0000084b28b2a80e in _libc_abort () at
> > /usr/src/lib/libc/stdlib/abort.c:51
> > #2 0x00000848adda823a in std::sys::unix::abort_internal ()
> > #3 0x00000848adda3d4a in std::process::abort ()
> > #4 0x00000848add97461 in rust_oom ()
> > #5 0x00000848addcbb6a in __rg_oom ()
> > #6 0x00000848addcd5ea in alloc::alloc::handle_alloc_error::rt_error ()
> > #7 0x00000848addcd5ba in core::ops::function::FnOnce::call_once ()
> > #8 0x00000848addcd5a9 in core::intrinsics::const_eval_select ()
> > #9 0x00000848addcd5d9 in alloc::alloc::handle_alloc_error ()
> > #10 0x00000848ad8299e6 in wu_diff::diff ()
> > #11 0x00000848ad84bfe4 in difft::diff::myers_diff::slice ()
> > #12 0x00000848ad831cc7 in difft::line_parser::change_positions ()
> > #13 0x00000848ad8349f0 in difft::diff_file ()
> > #14 0x00000848ad8335bc in difft::main ()
> > #15 0x00000848ad806e16 in
> > std::sys_common::backtrace::__rust_begin_short_backtrace ()
> > #16 0x00000848ad821d1c in std::rt::lang_start::{{closure}} ()
> > #17 0x00000848add96bf6 in std::rt::lang_start_internal ()
> > #18 0x00000848ad83dc7a in main ()
> >
> > On 2022/10/15 11:47, Volker Schlecht wrote:
> > > Bump.
> > >
> > > On 10/7/22 12:09, Volker Schlecht wrote:
> > > >
> > > >
> > > > On 10/7/22 10:02, Landry Breuil wrote:
> > > > > anyway, my port for 0.35 was much similar to yours, the only
> > > > > "relevant"
> > > > > differences i had were:
> > > > > SEPARATE_BUILD = Yes
> > > > > WANTLIB += c ${COMPILER_LIBCXX} m pthread
> > > > >
> > > > > (im never sure about the COMPILER_LIBCXX bit for rust ports..)
> > > >
> > > > I went for c++ and c++abi mostly because I would have had to explicitly
> > > > set COMPILER to shut up portcheck, so I chose your textproc/delta port
> > > > as an orientation.
> > > >
> > > > Regarding SEPARATE_BUILD: Is there anything particular about non-amd64
> > > > platforms needing that? It definitely isn't needed on amd64 here.
> > > >
> > > > > and DESCR:
> > > > > -Difftastic is an experimental diff tool that compares files based
> > > > > -on their syntax.
> > > > > -It supports over 30 programming languages.
> > > > > +Difftastic is a structural diff tool that understands syntax. It
> > > > > +supports over 20 programming languages, parses the code, and then
> > > > > +compares the syntax trees.
> > > > >
> > > > > ok landry@ to import your version, maybe just improve the DESCR and
> > > > > lowercase the first letter of COMMENT.
> > > >
> > > > Done, updated port attached!
> > > >
> > > > cu,
> > > > Volker
> > >
> >
