Much later than intended, I have finally produced a new release of iftop: 1.0pre3.
There were quite a few issues thrown up in pre2, hence a further pre-release.
This release is primarily bug fixes and build issues, but I've also added the extremely long-standing text mode output patch, which the mailing list ate when it was originally submitted in 2007.
Thanks to everyone who has contributed patches. Please let me know if you have any issues with this release, as if not I'll be calling it 1.0.
The release is here:
http://www.ex-parrot.com/~pdw/iftop/download/iftop-1.0pre3.tar.gz
Happy New Year!
Paul
Hi Paul,
Am 01.01.2014 um 20:14 schrieb Paul Warren pdw@ex-parrot.com:
Much later than intended, I have finally produced a new release of iftop: 1.0pre3.
There were quite a few issues thrown up in pre2, hence a further pre-release.
This release is primarily bug fixes and build issues, but I've also added the extremely long-standing text mode output patch, which the mailing list ate when it was originally submitted in 2007.
Thanks to everyone who has contributed patches. Please let me know if you have any issues with this release, as if not I'll be calling it 1.0.
The release is here:
http://www.ex-parrot.com/~pdw/iftop/download/iftop-1.0pre3.tar.gz
I am trying to use iftop on Solaris and found some issues:
1. I have ares available as libcares from http://c-ares.haxx.se For this to work I added cares as library to look for in configure.in I also needed to use AC_CONFIG_HEADERS or my autoreconf didn’t work with a message AM_CONFIG_HEADER being deprecated. http://sourceforge.net/p/gar/code/22770/tree//csw/mgar/pkg/iftop/trunk/files...
2. In Solaris SIOCGIFFLAGS is in <sys/sockio.h> which needs to be included. Probably a check in configure would be better than my hack from here: http://sourceforge.net/p/gar/code/22770/tree//csw/mgar/pkg/iftop/trunk/files...
3. Nonetheless I don’t get interface statistics to be displayed other than „0“. This also has been reported in a long-standing bug report: https://www.opencsw.org/mantis/view.php?id=4557 Do you have an advice how I can further track the issue? I can offer an account on our buildfarm if that helps, lots of other projects also do: http://www.opencsw.org/extend-it/signup/to-upstream-maintainers/
Best regards
— Dago
Hi Dago,
On 01/01/2014 20:34, Dagobert Michelsen wrote:
I am trying to use iftop on Solaris and found some issues:
- I have ares available as libcares from http://c-ares.haxx.se For this to work I added cares as library to look for in configure.in I also needed to use AC_CONFIG_HEADERS or my autoreconf didn’t work with a message AM_CONFIG_HEADER being deprecated. http://sourceforge.net/p/gar/code/22770/tree//csw/mgar/pkg/iftop/trunk/files...
OK - I'll merge that one in.
- In Solaris SIOCGIFFLAGS is in <sys/sockio.h> which needs to be included. Probably a check in configure would be better than my hack from here: http://sourceforge.net/p/gar/code/22770/tree//csw/mgar/pkg/iftop/trunk/files...
Ah yes, I think I've seen that before, but was reluctant to put it in as-is in case it breaks other platforms. As you say, autoconf magic is needed.
- Nonetheless I don’t get interface statistics to be displayed other than „0“. This also has been reported in a long-standing bug report: https://www.opencsw.org/mantis/view.php?id=4557 Do you have an advice how I can further track the issue? I can offer an account on our buildfarm if that helps, lots of other projects also do: http://www.opencsw.org/extend-it/signup/to-upstream-maintainers/
That might be helpful, although I don't know if iftop needs root to run effectively on Solaris?
In terms of tracking the issue, we need to figure out if pcap is giving us the packets and we're erroneously junking them, or if it's simply not giving us the packets.
Adding some debug at the very beginning of the relevant packet handler (handle_*_packet() in iftop.c. You can add some debug to packet_init to find out which packet handler it's actually using.
Paul
Hi Paul,
Am 02.01.2014 um 00:09 schrieb Paul Warren pdw@ex-parrot.com:
- In Solaris SIOCGIFFLAGS is in <sys/sockio.h> which needs to be included. Probably a check in configure would be better than my hack from here: http://sourceforge.net/p/gar/code/22770/tree//csw/mgar/pkg/iftop/trunk/files...
Ah yes, I think I've seen that before, but was reluctant to put it in as-is in case it breaks other platforms. As you say, autoconf magic is needed.
It looks like this one hasn’t made it to 1.0pre4, could you please add it before release?
Best regards
— Dago
Hi Paul,
On 01.01.2014 20:14, Paul Warren wrote: [...]
Please let me know if you have any issues with this release, as if not I'll be calling it 1.0.
I have tested the latest release and so far I have found only some small issues.
- The file configure.in is deprecated and upstream now recommends to use configure.ac instead. As far I know simply renaming the file should do the trick. See also https://lists.gnu.org/archive/html/automake/2013-05/msg00049.html for more information. - The new -t option is not mentioned in iftop.8. I have written a small patch for the man page. - There are several compiler warnings from gcc 4.8. Most of them are rather cosmetic issues and easy to fix by removing unused variables/functions or adding prototypes to header files. gcc-warnings.txt contains a list of those warnings. The gcc-warnings.patch fixes
warning: suggest parentheses around '&&' within '||' [-Wparentheses]
for example.
Do you have an estimated release date for 1.0? I could also upload ~pre3 to Debian unstable first and let a wider audience test this new release. If there were no major complaints, let's say within 4 weeks, 1.0 would be ready for a release.
Regards,
Markus
Hi Markus,
Thanks for the feedback.
On 05/01/2014 13:04, Markus Koschany wrote:
I have tested the latest release and so far I have found only some small issues.
- The file configure.in is deprecated and upstream now recommends to use configure.ac instead. As far I know simply renaming the file should do the trick. See also https://lists.gnu.org/archive/html/automake/2013-05/msg00049.html for more information.
I'll take a look at doing that. I always find autoconf/automake a painful black art, but hopefully this will be straightforward.
- The new -t option is not mentioned in iftop.8. I have written a small patch for the man page.
Thank you.
There are several compiler warnings from gcc 4.8. Most of them are rather cosmetic issues and easy to fix by removing unused variables/functions or adding prototypes to header files. gcc-warnings.txt contains a list of those warnings. The gcc-warnings.patch fixes
warning: suggest parentheses around '&&' within '||' [-Wparentheses]
for example.
Someone else had already mentioned that - thanks for the patch.
I'll review the other warnings.
Do you have an estimated release date for 1.0? I could also upload ~pre3 to Debian unstable first and let a wider audience test this new release. If there were no major complaints, let's say within 4 weeks, 1.0 would be ready for a release.
No specific date other than ASAP. There's enough come up in pre3 to warrant a pre4, which will then hopefully be released unchanged as 1.0.
I would expect any issues introduced in pre3 to be build-time rather than run time, and I've now had good feedback on that. The introduction of the "-t" mode was more of a change than I was planning to make at this stage, but I'm pretty confident that it won't result in regressions in any functionality that was present in pre2.
Hopefully I'll get a pre4 out today, and if we're happy that it builds OK on all platforms that earlier releases have then I'll press the button on 1.0 within a week or so.
Paul