Btw, if you have a succesful port, we could already apply the diff to Debian and get people to test the result, before we submit to upstream. Just ping me when you're ready.
Martin-Éric ma 14.10.2024 klo 13.32 Martin-Éric Racine (martin-eric.rac...@iki.fi) kirjoitti: > > Hey Joan, > > Thanks for looking into this. If you ever succeed in porting dhcpcd to > Hurd, please submit a diff to upstream. He will gladly merge it. > > Martin-Éric > > pe 11. lokak. 2024 klo 6.59 Damien Zammit (dam...@zamaudio.com) kirjoitti: > > > > I'm guessing its because of routes being defined as ioctls with slightly > > different header files in Hurd vs BSD (?) Likely Samuel's solution is > > correct to guard against incompatible headers being included in the Hurd > > path. > > Alternatively one could group the Hurd headers better so both APIs require > > to include the same filenames in both paths. > > > > That's my guess. > > > > Damien > > > > Sent from Proton Mail mobile > > > > > > > > -------- Original Message -------- > > On 11 Oct 2024, 2:05 am, Joan Lledó < jlle...@mailfence.com> wrote: > > > > > > Hi, This comes from: > > https://lists.debian.org/debian-hurd/2023/11/msg00030.html I was trying to > > build the dhcpcd package on the Hurd and it fails with this error: ``` In > > file included from arp.h:46, from privsep.h:134, from dhcpcd.h:96, from > > common.c:40: if.h:36:10: fatal error: netinet/in_var.h: No such file or > > directory 36 | #include /* for IN_IFF_TENTATIVE et all */ | > > ^~~~~~~~~~~~~~~~~~ ``` Source: > > https://salsa.debian.org/debian/dhcpcd/-/blob/upstream/10.1.0/src/if.h?ref_type=tags#L36 > > It fails because apparently we define the `BSD` macro. I checked it and it > > comes from glibc: > > https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/mach/hurd/bits/param.h;h=48eeb7ddf5d62942d15705e0bcc19db05ed28fc7;hb=HEAD#l63 > > So, according to the comment above the macro definition, we claim to be > > compatible with BSD, is that still true? dchpcd reads the `BSD` macro from > > multiple places across its codebase. If it fails to compile, it's because > > we are not compatible apparently. How can we fix this?
