spiped 1.3.0 coming soon

[Colin Percival]

Hi all,

I'm hoping to release spiped-1.3.0 in the next few days. Non-trivial changes
since 1.2.2 are:

* Man pages are included (thanks Andreas!) and installed if you specify MAN1DIR=
on the 'make install' command line.

* If asprintf is a macro it's undefined before being redefined.

* _POSIX_C_SOURCE is set to 200809L.

* Protocol-layer keep-alives are enabled, unless -j is specified, in which case
they are disabled.

* In spiped, the target address is re-resolved every 60 seconds, unless -R is
specified, which disables address re-resolution, or -r <rtime> is specified,
which changes the time between address re-resolutions to <rtime> seconds.

Source code is at https://code.google.com/p/spiped/.

Is there anything else anyone thinks ought to be included in 1.3.0?

[James Turner]

Post by Colin Percival
Hi all,
I'm hoping to release spiped-1.3.0 in the next few days. Non-trivial changes
* Man pages are included (thanks Andreas!) and installed if you specify MAN1DIR=
on the 'make install' command line.
* If asprintf is a macro it's undefined before being redefined.
* _POSIX_C_SOURCE is set to 200809L.
* Protocol-layer keep-alives are enabled, unless -j is specified, in which case
they are disabled.
* In spiped, the target address is re-resolved every 60 seconds, unless -R is
specified, which disables address re-resolution, or -r <rtime> is specified,
which changes the time between address re-resolutions to <rtime> seconds.
Source code is at https://code.google.com/p/spiped/.
Is there anything else anyone thinks ought to be included in 1.3.0?
--
Colin Percival
Security Officer Emeritus, FreeBSD | The power to serve
Founder, Tarsnap | www.tarsnap.com | Online backups for the truly paranoid

So I checked out spiped from svn then ran VERSION=1.3.0 make publish.
I then tested building from there on an OpenBSD amd64 snapshot from 4/1.
With the new addition of 200809L anywhere sys/sockets.h was added I also
had to add sys/types.h.

Now that spiped uses dnsthread which in turn uses pthread functions I
had to add -lpthread to the list of spiped LDADD_REQs.

Diff attached. Hopefully this was also the sort of thing you were
looking for.

[Colin Percival]

Post by James Turner
So I checked out spiped from svn then ran VERSION=1.3.0 make publish.
I then tested building from there on an OpenBSD amd64 snapshot from 4/1.
With the new addition of 200809L anywhere sys/sockets.h was added I also
had to add sys/types.h.

It seems that guenther's 1.82 in sys/sys/socket.h fixed the <sys/types.h>
inclusion but only with __BSD_VISIBLE, which is disabled due to spiped
having _POSIX_C_SOURCE set. What's the precise error you see? Perhaps
OpenBSD needs more #ifndef/#define/typedef/#endif clauses in socket.h
like it already has for socklen_t, sa_family_t, size_t, and ssize_t to
handle the !__BSD_VISIBLE case?

Post by James Turner
Now that spiped uses dnsthread which in turn uses pthread functions I
had to add -lpthread to the list of spiped LDADD_REQs.

Thanks, fixed.

Post by James Turner
Diff attached. Hopefully this was also the sort of thing you were
looking for.

Exactly the sort of thing I was looking for.

[James Turner]

Post by Colin Percival

Post by James Turner
So I checked out spiped from svn then ran VERSION=1.3.0 make publish.
I then tested building from there on an OpenBSD amd64 snapshot from 4/1.
With the new addition of 200809L anywhere sys/sockets.h was added I also
had to add sys/types.h.

It seems that guenther's 1.82 in sys/sys/socket.h fixed the <sys/types.h>
inclusion but only with __BSD_VISIBLE, which is disabled due to spiped
having _POSIX_C_SOURCE set. What's the precise error you see? Perhaps
OpenBSD needs more #ifndef/#define/typedef/#endif clauses in socket.h
like it already has for socklen_t, sa_family_t, size_t, and ssize_t to
handle the !__BSD_VISIBLE case?

/usr/include/sys/socket.h:189: error: expected specifier-qualifier-list
before 'u_int8_t'
/usr/include/sys/socket.h:207: error: expected specifier-qualifier-list
before 'u_int8_t'

Post by Colin Percival

Post by James Turner
Now that spiped uses dnsthread which in turn uses pthread functions I
had to add -lpthread to the list of spiped LDADD_REQs.

Thanks, fixed.

Post by James Turner
Diff attached. Hopefully this was also the sort of thing you were
looking for.

Exactly the sort of thing I was looking for.
--
Colin Percival
Security Officer Emeritus, FreeBSD | The power to serve
Founder, Tarsnap | www.tarsnap.com | Online backups for the truly paranoid

[Colin Percival]

Post by James Turner

Post by Colin Percival

Post by James Turner
So I checked out spiped from svn then ran VERSION=1.3.0 make publish.
I then tested building from there on an OpenBSD amd64 snapshot from 4/1.
With the new addition of 200809L anywhere sys/sockets.h was added I also
had to add sys/types.h.

It seems that guenther's 1.82 in sys/sys/socket.h fixed the <sys/types.h>
inclusion but only with __BSD_VISIBLE, which is disabled due to spiped
having _POSIX_C_SOURCE set. What's the precise error you see? Perhaps
OpenBSD needs more #ifndef/#define/typedef/#endif clauses in socket.h
like it already has for socklen_t, sa_family_t, size_t, and ssize_t to
handle the !__BSD_VISIBLE case?

/usr/include/sys/socket.h:189: error: expected specifier-qualifier-list
before 'u_int8_t'
/usr/include/sys/socket.h:207: error: expected specifier-qualifier-list
before 'u_int8_t'

Looks to me like you need to apply s/u_int/__uint/g to sys/socket.h -- can
you try that and see if the errors go away?

[James Turner]

Post by Colin Percival

Post by James Turner

Post by Colin Percival

Post by James Turner
So I checked out spiped from svn then ran VERSION=1.3.0 make publish.
I then tested building from there on an OpenBSD amd64 snapshot from 4/1.
With the new addition of 200809L anywhere sys/sockets.h was added I also
had to add sys/types.h.

It seems that guenther's 1.82 in sys/sys/socket.h fixed the <sys/types.h>
inclusion but only with __BSD_VISIBLE, which is disabled due to spiped
having _POSIX_C_SOURCE set. What's the precise error you see? Perhaps
OpenBSD needs more #ifndef/#define/typedef/#endif clauses in socket.h
like it already has for socklen_t, sa_family_t, size_t, and ssize_t to
handle the !__BSD_VISIBLE case?

/usr/include/sys/socket.h:189: error: expected specifier-qualifier-list
before 'u_int8_t'
/usr/include/sys/socket.h:207: error: expected specifier-qualifier-list
before 'u_int8_t'

Looks to me like you need to apply s/u_int/__uint/g to sys/socket.h -- can
you try that and see if the errors go away?
--
Colin Percival
Security Officer Emeritus, FreeBSD | The power to serve
Founder, Tarsnap | www.tarsnap.com | Online backups for the truly paranoid

Yup, that fixed the problem. socket.h happens to reference u_int8_t
twice and u_int64_t once. I'll see about getting them changed to __uint
instead.

[Colin Percival]

Post by James Turner

Post by Colin Percival
Looks to me like you need to apply s/u_int/__uint/g to sys/socket.h -- can
you try that and see if the errors go away?

Yup, that fixed the problem. socket.h happens to reference u_int8_t
twice and u_int64_t once. I'll see about getting them changed to __uint
instead.

Thanks! Let me know when that's done and I'll update the BUILDING file to
say which versions of OpenBSD are affected.

[James Turner]

Post by Colin Percival

Post by James Turner

Post by Colin Percival
Looks to me like you need to apply s/u_int/__uint/g to sys/socket.h -- can
you try that and see if the errors go away?

Yup, that fixed the problem. socket.h happens to reference u_int8_t
twice and u_int64_t once. I'll see about getting them changed to __uint
instead.

Thanks! Let me know when that's done and I'll update the BUILDING file to
say which versions of OpenBSD are affected.
--
Colin Percival
Security Officer Emeritus, FreeBSD | The power to serve
Founder, Tarsnap | www.tarsnap.com | Online backups for the truly paranoid

Looks like you saw the commit go in. This will effect OpenBSD 5.4 users
and beyond since 5.3 has already been tagged and will be released May
1st. Or if you use snapshots it will be in any snapshot dated after
April 2nd.

[Colin Percival]

Hi all,

In case it helps encourage people to test, I've attached a tarball to this
email so you don't need to worry about checking out the source from SVN.

For clarity, I'm interested in both reports of build breakage and also any
minor changes / improvements people would like to see in spiped (realizing
that if you don't get it in now, you probably won't see it for at least 6
months).

Colin Percival

Post by Colin Percival
Hi all,
I'm hoping to release spiped-1.3.0 in the next few days. Non-trivial changes
* Man pages are included (thanks Andreas!) and installed if you specify MAN1DIR=
on the 'make install' command line.
* If asprintf is a macro it's undefined before being redefined.
* _POSIX_C_SOURCE is set to 200809L.
* Protocol-layer keep-alives are enabled, unless -j is specified, in which case
they are disabled.
* In spiped, the target address is re-resolved every 60 seconds, unless -R is
specified, which disables address re-resolution, or -r <rtime> is specified,
which changes the time between address re-resolutions to <rtime> seconds.
Source code is at https://code.google.com/p/spiped/.
Is there anything else anyone thinks ought to be included in 1.3.0?