Re: [Monetdb-developers] ! Testing 'Current' sql: Results DIFFER from 2007.04.28_04-03-41's (GDKdebug=10) !
Project account wrote:
Testing results of sql in /export/scratch2/monet/Testing/Current with GDKdebug=10 DIFFER from those of 2007.04.28_04-03-41 !
Maybe YOUR checkins since 28.04.2007, 04:03:41 caused some of the problems:
buildtools MonetDB clients MonetDB4 MonetDB5 mk@cwi.nl sql mk@cwi.nl niels@cwi.nl romulog@users.sourceforge.net
Check http://monetdb.cwi.nl/Development/TestWeb/Current/sql/ for details !
Old file: Mtest10..2007.04.28_04-03-41..out New file: Mtest10..2007.04.29_04-16-39..out ===========================================
Summary:
- 1 New systems
!! 3 Systems now DON'T produce output !!
I compiled my sources before the commit with optimization on: --enable-strict --enable-assert --enable-optimize --enable-bits=64
Why the variable initialized, but never used is not detected by this compilation? Which compilation must be done to detect these kind of errors?
Regards, Romulo
Romulo,
you most probably compiled with gcc (the default), right. That's fine. Moreover, gcc also works fine with testing.
However, icc (the Intel compiler) is a bit more "picky" (read: *accurate*) and points out that in function create_func(), variable "sub_f" is set, but never used --- looking at the code, I must say icc is (as always) very right so:
In fucntion create_func() in sql/src/server/sql_psm.mx, variable "sub_f" is declared and initialized (set) to NULL in line 508. Then, in line 525, variable "sub_f" is set to "sql_bind_func_(s, fname, type_list)". However, there is no further occurance of variable "sub_f" in function create_func(); hence, it is set (see above) but *never* *used*...
... either you "forgot" to used (the content of) variable "sub_f" "somewhere" in function create_func(), or you don't need it. In the former case, icc helped you to find a BUG in your code. In the latter case, you should remove variable "sub_f" from function create_func(), i.e., remove line 508 and change line 525 from if (!instantiate && (sub_f = sql_bind_func_(s, fname, type_list))) { to if (!instantiate && sql_bind_func_(s, fname, type_list)) {
Stefan
On Sun, Apr 29, 2007 at 11:57:03AM +0200, Romulo Goncalves wrote:
Project account wrote:
Testing results of sql in /export/scratch2/monet/Testing/Current with GDKdebug=10 DIFFER from those of 2007.04.28_04-03-41 !
Maybe YOUR checkins since 28.04.2007, 04:03:41 caused some of the problems:
buildtools MonetDB clients MonetDB4 MonetDB5 mk@cwi.nl sql mk@cwi.nl niels@cwi.nl romulog@users.sourceforge.net
Check http://monetdb.cwi.nl/Development/TestWeb/Current/sql/ for details !
Old file: Mtest10..2007.04.28_04-03-41..out New file: Mtest10..2007.04.29_04-16-39..out ===========================================
Summary:
- 1 New systems
!! 3 Systems now DON'T produce output !!
I compiled my sources before the commit with optimization on: --enable-strict --enable-assert --enable-optimize --enable-bits=64
Why the variable initialized, but never used is not detected by this compilation? Which compilation must be done to detect these kind of errors?
Regards, Romulo
This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Monetdb-developers mailing list Monetdb-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/monetdb-developers
Stefan Manegold wrote:
Romulo,
you most probably compiled with gcc (the default), right. That's fine. Moreover, gcc also works fine with testing.
However, icc (the Intel compiler) is a bit more "picky" (read: *accurate*) and points out that in function create_func(), variable "sub_f" is set, but never used --- looking at the code, I must say icc is (as always) very right so:
In fucntion create_func() in sql/src/server/sql_psm.mx, variable "sub_f" is declared and initialized (set) to NULL in line 508. Then, in line 525, variable "sub_f" is set to "sql_bind_func_(s, fname, type_list)". However, there is no further occurance of variable "sub_f" in function create_func(); hence, it is set (see above) but *never* *used*...
... either you "forgot" to used (the content of) variable "sub_f" "somewhere" in function create_func(), or you don't need it. In the former case, icc helped you to find a BUG in your code. In the latter case, you should remove variable "sub_f" from function create_func(), i.e., remove line 508 and change line 525 from if (!instantiate && (sub_f = sql_bind_func_(s, fname, type_list))) { to if (!instantiate && sql_bind_func_(s, fname, type_list)) {
Yes I know, I just like to detect these things before my commit, but if the compiler is icc I prefer to wait for the compilation results during the night ;)
The fix is already check in.
Regards, Romulo
Stefan
On Sun, Apr 29, 2007 at 11:57:03AM +0200, Romulo Goncalves wrote:
Project account wrote:
Testing results of sql in /export/scratch2/monet/Testing/Current with GDKdebug=10 DIFFER from those of 2007.04.28_04-03-41 !
Maybe YOUR checkins since 28.04.2007, 04:03:41 caused some of the problems:
buildtools MonetDB clients MonetDB4 MonetDB5 mk@cwi.nl sql mk@cwi.nl niels@cwi.nl romulog@users.sourceforge.net
Check http://monetdb.cwi.nl/Development/TestWeb/Current/sql/ for details !
Old file: Mtest10..2007.04.28_04-03-41..out New file: Mtest10..2007.04.29_04-16-39..out ===========================================
Summary:
- 1 New systems
!! 3 Systems now DON'T produce output !!
I compiled my sources before the commit with optimization on: --enable-strict --enable-assert --enable-optimize --enable-bits=64
Why the variable initialized, but never used is not detected by this compilation? Which compilation must be done to detect these kind of errors?
Regards, Romulo
This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Monetdb-developers mailing list Monetdb-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/monetdb-developers
Hi Romulo,
I had the same problem as you some time ago. gcc is by default less picky than icc:( I think you can add the option "--enable-strict" to 'configure', which should make your gcc more strict.
Regards,
Jennie
On Sun, Apr 29, 2007 at 12:14:55PM +0200, Romulo Goncalves wrote:
Stefan Manegold wrote:
Romulo,
you most probably compiled with gcc (the default), right. That's fine. Moreover, gcc also works fine with testing.
However, icc (the Intel compiler) is a bit more "picky" (read: *accurate*) and points out that in function create_func(), variable "sub_f" is set, but never used --- looking at the code, I must say icc is (as always) very right so:
In fucntion create_func() in sql/src/server/sql_psm.mx, variable "sub_f" is declared and initialized (set) to NULL in line 508. Then, in line 525, variable "sub_f" is set to "sql_bind_func_(s, fname, type_list)". However, there is no further occurance of variable "sub_f" in function create_func(); hence, it is set (see above) but *never* *used*...
... either you "forgot" to used (the content of) variable "sub_f" "somewhere" in function create_func(), or you don't need it. In the former case, icc helped you to find a BUG in your code. In the latter case, you should remove variable "sub_f" from function create_func(), i.e., remove line 508 and change line 525 from if (!instantiate && (sub_f = sql_bind_func_(s, fname, type_list))) { to if (!instantiate && sql_bind_func_(s, fname, type_list)) {
Yes I know, I just like to detect these things before my commit, but if the compiler is icc I prefer to wait for the compilation results during the night ;)
The fix is already check in.
Regards, Romulo
Stefan
On Sun, Apr 29, 2007 at 11:57:03AM +0200, Romulo Goncalves wrote:
Project account wrote:
Testing results of sql in /export/scratch2/monet/Testing/Current with GDKdebug=10 DIFFER from those of 2007.04.28_04-03-41 !
Maybe YOUR checkins since 28.04.2007, 04:03:41 caused some of the problems:
buildtools MonetDB clients MonetDB4 MonetDB5 mk@cwi.nl sql mk@cwi.nl niels@cwi.nl romulog@users.sourceforge.net
Check http://monetdb.cwi.nl/Development/TestWeb/Current/sql/ for details !
Old file: Mtest10..2007.04.28_04-03-41..out New file: Mtest10..2007.04.29_04-16-39..out ===========================================
Summary:
- 1 New systems
!! 3 Systems now DON'T produce output !!
I compiled my sources before the commit with optimization on: --enable-strict --enable-assert --enable-optimize --enable-bits=64
Why the variable initialized, but never used is not detected by this compilation? Which compilation must be done to detect these kind of errors?
Regards, Romulo
This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Monetdb-developers mailing list Monetdb-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/monetdb-developers
This SF.net email is sponsored by DB2 Express Download DB2 Express C - the FREE version of DB2 express and take control of your XML. No limits. Just data. Click to get it now. http://sourceforge.net/powerbar/db2/ _______________________________________________ Monetdb-developers mailing list Monetdb-developers@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/monetdb-developers
participants (3)
-
Romulo Goncalves -
Stefan Manegold -
Ying Zhang