Theo Buehler <t...@theobuehler.org> wrote:
> On Thu, Jul 21, 2022 at 01:55:41PM +0200, Omar Polo wrote:
> > Stuart Henderson <s...@spacehopper.org> wrote:
> > > On 2022/07/20 17:15, tux0r wrote:
> > > > Stuart Henderson schrob am 20.07.2022 um 14:21:
> > > > 
> > > > >The COMMENT assumes that one knows what ZPAQ is. It would be better if
> > > > >it described what the software does, and leave "it is a fork of foo" to
> > > > >DESCR.
> > > > 
> > > > Makes sense. Attached is an updated archive. Having ZPAQ in the COMMENT 
> > > > still sounds reasonable to me, as it is a (relatively) well-known 
> > > > software and people might search for it directly.
> > > 
> > > There's a size limit on COMMENT, see bsd.port.mk(5) for details.
> > > I propose "journaling archiver for incremental backup, fork of ZPAQ"
> > > which fits. (No need to send a new tar).
> > > 
> > > With that change it's OK sthen@ to import
> > 
> > imported, thanks!
> 
> The build with -Dunix -DNOJIT is broken:
>   -DNOJIT removes #include <sys/mman.h>

the patch for it was lost in some iteration, sorry.  i even sent a pr
upstream[0], but apparently upstream overwrote it in a follow-up commit
(one of the "Add files via upload" commits...)

i should have checked again before importing, apologize.

[0]: https://github.com/fcorbelli/zpaqfranz/pull/22

>   -Dunix makes use of RWX mappings unconditionally.
> 
> The forcelatinansi function is broken on unsigned char architectures,
> which means that this won't work as expected with -utf or -utf8:
> 
> zpaqfranz.cpp:19226:19: warning: result of comparison of constant -1 with 
> expression of type 'std::basic_string<char>::value_type' (aka 'char') is 
> always false [-Wtautological-constant-out-of-range-compare]
>                         if (i_string[j]==-1) // °
> 
> http://build-failures.rhaalovely.net/aarch64/2022-07-22/archivers/zpaqfranz.log
> 
> 
> I could not come up with a non-stupid fix of this in 5 minutes, so I
> left it as it is.
> 
> Diff below is untested, but at least it makes the port build on !amd64.

I came up with the conditional because i thought it wouldn't build on
!amd64 without -DNOJIT.  i misunderstood some bits of the jit probably.
anyway, it still builds and works on amd64, if it compiles on other
arches too ok op@

Thanks!

Reply via email to