[Pianod] Build 111 - user persistence changes

Peter Li chinasaurli at gmail.com
Tue Feb 12 16:20:00 PST 2013


It looks like the (pianod) Makefile still needs to be updated with 
-lcrypt if __linux__

Unfortunately I started getting that segfault from the call to crypt() 
again.  Only on my FC machine though, not on ALARM or Raspbian.  Seems 
like #include <crypt.h> does not necessarily get it to 
/usr/include/crypt.h instead of .../src/libpiano/crypt.h.

The only way to get the segfault to go away on my current system was to 
rename libpiano/crypt.h to libpiano/libpianocrypt.h.  This of course 
meant also changing: libpiano/crypt.c libpiano/request libpiano/response 
and the Makefile.

I can send you a patch if you want to go this way.  Or maybe there's 
someone upstream at libpiano that I can talk to about not calling their 
file "crypt.h"?  Or is there a better fix that will get my system to 
distinguish <crypt.h> from "crypt.h" the way I thought it should?  Or do 
you want to punt and change encrypt_password to use a different method?

Sorry...
P


On 02/02/2013 01:47 PM, Peter Li wrote:
> Okay I see: crypt is a small standard lib that v111 uses alongside 
> gcrypt.
>
> So I needed to add -lcrypt to the link options.  Then it builds but 
> segfaults, perhaps because it gets confused by the crypt.h that is 
> inside libpiano?
>
> So if you remove the implicit declaration from users.c by putting 
> #include <crypt.h> in then it seems to run okay.
>
> Can send you a patch but it's pretty clear from above I think?
> P
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.deviousfish.com/pipermail/pianod-deviousfish.com/attachments/20130212/24a4bbaf/attachment-0003.htm>


More information about the Pianod mailing list