rtptools: move to newer github

[mitchblank]

It seems that "columbia-irt" now is using "irtlab" github for this project.

The homepage with its tarballs is still gone, but you can get tarballs from the new github. However these don't include ./configure so we need to run autoconf before building

[mitchblank]

Huh, this is annoying... I swear I tried a brew livecheck before submitting...

[nandahkrishna]

The repository isn't tagging their latest release, so we shouldn't be using the latest URL here. We can simply use

livecheck do
  url :head
end

This will check the git tags instead. I'll add a suggestion.

[mitchblank]

This will check the git tags instead.

Thanks. Sorry for the bother, it's my first time trying to mess with livecheck

[nandahkrishna]

No issues, haha. I'm working on some changes that should make this process easier 👍🏼

[fxcoudert]

Is this really the same upstream project? Or a fork

[mitchblank]

I have no first-hand knowledge but https://github.com/irtlab claims to be the github of the group at Columbia that created this library. Assuming that's not completely fraudulent it seems like it's the closest place to an official repo of the sources. The https://github.com/columbia-irt account seems to be (more) abandoned

Doesn't look like this particular library is under and development at this time though; it doesn't seem to have a live presence on IRT's webpage.

Perhaps @mcd500, @janstary, or @sobomax can clear up this question more since they seem to be the recent committers to that repo and all remain active on github.

[carlocab]

Perhaps it might help to file an issue to ask for a clarification?

[mitchblank]

Perhaps it might help to file an issue to ask for a clarification?

There already is a similar irtlab/rtptools#135 open since June (look at the thread) with no answer.

[carlocab]

Oof. Ok. Do other package managers ship rtptools? Have any of them switched to github for their source?

[mitchblank]

Oof. Ok. Do other package managers ship rtptools? Have any of them switched to github for their source?

It looks like archilinux does https://aur.archlinux.org/packages/rtptools However they're still pointing at the 404 links at www.cs.columbia.edu so they don't have a real upstream at the minute.

[carlocab]

Yea, doesn't look like anyone else has switched:

https://repology.org/project/rtptools/versions

Maybe the formula should just be deprecated? It's not maintained, had 101 installs in the past year... It doesn't look great.

[mitchblank]

Maybe the formula should just be deprecated?

I don't care one way or another. I only noticed this formula because it's missing a Big Sur bundle and I tried to fix it 🤷

Of course if we deprecate-but-dont-disable that still leaves the question of fixing the compilation.

[mitchblank]

Out of curiosity, I downloaded the 1.22 tarball from http://distfiles.macports.org/rtptools/ (which matches the "2c76b2a423fb943820c91194372133a44cbdc456ebf69c51616ec50eeb068c28" of the source tarball we used to track before our source stopped working) and compared it versus the new github-hosted one.

As I mentioned before, the big change is a difference it what build scripts are included:

$ diff -urN from-macports new-github | diffstat
 .gitignore       |   19
 Makefile.in      |  953 ---------------------
 aclocal.m4       | 1152 -------------------------
 ansi.h           |   17
 compile          |  347 -------
 config.guess     | 1441 --------------------------------
 config.sub       | 1813 ----------------------------------------
 configure        | 6775 --------------------------------------------------------------------------------------------------------------------------------------------------------
 depcomp          |  791 -----------------
 install-sh       |  508 -----------
 missing          |  215 ----
 rtpdump.c        |    1
 win/rtpdump.suo  |binary
 win/rtpplay.suo  |binary
 win/rtpsend.suo  |binary
 win/rtptrans.suo |binary
 16 files changed, 20 insertions(+), 14012 deletions(-)

there are some minimal changes in ansi.h and rtpdump.h but they're just related to when to #include <math.h> so nothing that looks nefarious:

--- from-macports/ansi.h·······┊2018-03-22 01:40:41.000000000 +0000
+++ new-github/ansi.h··┊2018-03-24 13:37:56.000000000 +0000
@@ -166,21 +166,11 @@
 extern int getdomainname(char *name, int namelen);
 #endif
●
-/* <math.h> */
-#ifdef _MATH_H
-extern double infinity(void);
-#endif
-
-#endif /* sun4 */
+#endif /* sun */
●
 /*** SGI ***/
 #ifdef sgi
●
-/* <math.h> */
-#ifdef __MATH_H__
-extern double infinity(void);
-#endif
-
 #ifdef __RPCSVC_YPCLNT_H__
 extern int yp_get_default_domain(char **outdomain);
 #endif
@@ -189,11 +179,6 @@
 /*** HP ***/
 #ifdef hp
●
-/* <math.h> */
-#ifdef _MATH_INCLUDED
-extern double infinity(void);
-#endif
-
 /* <unistd.h> */
 #ifdef _UNISTD_INCLUDED
 extern int getdomainname(char *name, int namelen);
diff -urN from-macports/rtpdump.c new-github/rtpdump.c
--- from-macports/rtpdump.c····┊2018-03-22 01:40:41.000000000 +0000
+++ new-github/rtpdump.c·······┊2018-03-24 13:37:56.000000000 +0000
@@ -39,7 +39,6 @@
 #include <errno.h>
 #include <unistd.h>      /* select(), perror() */
 #include <stdlib.h>      /* getopt(), atoi() */
-#include <math.h>        /* fmod() */
 #include <fcntl.h>       /* O_BINARY on Win32 */
 #include "rtp.h"
 #include "vat.h"

So the old and new tarball seem to be basically the same code, but probably just tagged slightly different or something.

[BrewTestBot]

This pull request has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.

[BrewTestBot]

@carlocab has triggered a merge.