Please, did you test it on older distributions or just on Fedora 32? Should we do a test on our older distributions?
In any case, there is required to use the Signed-off-by line in the commit message to declare consent to the DCO. Please, can you do an amend commit message and do a push using -f?
By the way, as I mentioned earlier, it is preferred to use the shared collaboration with branches model instead of creating unnecessary forks.
Please, did you test it on older distributions or just on Fedora 32? Should we do a test on our older distributions?
In any case, there is required to use the Signed-off-by line in the commit message to declare consent to the DCO. Please, can you do an amend commit message and do a push using -f?
By the way, as I mentioned earlier, it is preferred to use the shared collaboration with branches model instead of creating unnecessary forks.
Hello, I tested this patch only on Fedora 32, it is safer to test it on older distributions.
I've added the 'signed-off by' in the comment.
OK next time I will use a branch instead of fork.
Hello, I tested this patch only on Fedora 32, it is safer to test it on older distributions.
I've added the 'signed-off by' in the comment.
OK next time I will use a branch instead of fork.
This commit causes a regression as described in bug 3154. Reverting the commit fixed the regression, so I guess we need to find a different solution to bug 3120
This commit causes a regression as described in bug [3154](https://bugs.pearsoncomputing.net/show_bug.cgi?id=3154). Reverting the commit fixed the regression, so I guess we need to find a different solution to bug [3120](https://bugs.pearsoncomputing.net/show_bug.cgi?id=3120)
@Francois, could you check and confirm if reverting your commit still gives you FTBFS? In Debian we don't see the problem, so it is difficult for us to test a fix. In first place we need to make sure the problem is still there.
@Francois, could you check and confirm if reverting your commit still gives you FTBFS? In Debian we don't see the problem, so it is difficult for us to test a fix. In first place we need to make sure the problem is still there.
tdepim/libkholidays/scanholiday.c:
/* this should not be removed /
#define yyin kcalin
/ kcalin should be defined anywhere else */
extern FILE *yyin, yyout;
/ but kcalin is initialized here then defined here */
FILE *yyin = (FILE *) 0, *yyout = (FILE *) 0;
tdepim/libkholidays/parseholiday.c:
/* kcalin is here too */
FILE kcalin; / file currently being processed */
this multiple definition should be resolved instead.
tdepim/libkholidays/scanholiday.c:
/* this should not be removed */
#define yyin kcalin
/* kcalin should be defined anywhere else */
extern FILE *yyin, *yyout;
/* but kcalin is initialized here then defined here */
FILE *yyin = (FILE *) 0, *yyout = (FILE *) 0;
tdepim/libkholidays/parseholiday.c:
/* kcalin is here too */
FILE *kcalin; /* file currently being processed */
this multiple definition should be resolved instead.
Here's one more fact: You can see that scanholiday.lex and parseholiday.y are in the same directory. These are the source files from which scanholiday.c and parseholiday.{c,h} are generated using flex / bison. So here it seems like a good idea to make a modification to cmake rules so that these files are generated during the build, not part of the source code.
Here's one more fact: You can see that `scanholiday.lex` and `parseholiday.y` are in the same directory. These are the source files from which `scanholiday.c` and `parseholiday.{c,h}` are generated using flex / bison. So here it seems like a good idea to make a modification to cmake rules so that these files are generated during the build, not part of the source code.
sorry wrong description, this fix is for libkholiday build error
Fix kexi build with libpqxx 7σε Fix libkholiday build 4 έτη πρινPlease, did you test it on older distributions or just on Fedora 32? Should we do a test on our older distributions?
In any case, there is required to use the Signed-off-by line in the commit message to declare consent to the DCO. Please, can you do an amend commit message and do a push using -f?
By the way, as I mentioned earlier, it is preferred to use the shared collaboration with branches model instead of creating unnecessary forks.
Hello, I tested this patch only on Fedora 32, it is safer to test it on older distributions.
I've added the 'signed-off by' in the comment.
OK next time I will use a branch instead of fork.
Thanks Francois. Let us know after you have tested older Fedora distros, before we proceed with merging this 😄
Tested on centos6, centos7, mageia6, fedora31, no build issue with this patch.
Thanks Francois!
This resolves bug 3120
This commit causes a regression as described in bug 3154. Reverting the commit fixed the regression, so I guess we need to find a different solution to bug 3120
@Francois, could you check and confirm if reverting your commit still gives you FTBFS? In Debian we don't see the problem, so it is difficult for us to test a fix. In first place we need to make sure the problem is still there.
tdepim/libkholidays/scanholiday.c:
/* this should not be removed /
#define yyin kcalin
/ kcalin should be defined anywhere else */
extern FILE *yyin, yyout;
/ but kcalin is initialized here then defined here */
FILE *yyin = (FILE *) 0, *yyout = (FILE *) 0;
tdepim/libkholidays/parseholiday.c:
/* kcalin is here too */
FILE kcalin; / file currently being processed */
this multiple definition should be resolved instead.
Here's one more fact: You can see that
scanholiday.lexandparseholiday.yare in the same directory. These are the source files from whichscanholiday.candparseholiday.{c,h}are generated using flex / bison. So here it seems like a good idea to make a modification to cmake rules so that these files are generated during the build, not part of the source code.By generating those files using bison and lex I have been able to reproduce the error reported in bug 3120.
PR #46 provides a possible solution. Please help testing it on your system.
1e14bf4b2e.