Thread

1. scan.c:900: warning: ... postgresql-7.1

   Limin Liu <limin@pumpkinnet.com> — 2001-04-17T18:56:36Z

   Does anyone get the same warning messages from postgresql-7.1?
   
   It's under src/backend/parser...
   
   scan.c: In function `base_yylex':
   scan.c:900: warning: `yy_cp' might be used uninitialized in this
   function
   scan.c:900: warning: `yy_bp' might be used uninitialized in this
   function
   
   Here is the scan.c line 900
   --
   register char *yy_cp, *yy_bp;
   --
   
   In my previous postgres copy (7.1 beta 4), I did find the initialization
   for these two variables in scan.c.
   
   Can anyone please tell me why the init. is removed?  Thank you very
   much.
   
   
   --
   LM Liu
   
   
   
   

2. Re: scan.c:900: warning: ... postgresql-7.1

   Tom Lane <tgl@sss.pgh.pa.us> — 2001-04-17T20:36:44Z

   Limin Liu <limin@pumpkinnet.com> writes:
   > Does anyone get the same warning messages from postgresql-7.1?
   > It's under src/backend/parser...
   
   > scan.c: In function `base_yylex':
   > scan.c:900: warning: `yy_cp' might be used uninitialized in this
   > function
   > scan.c:900: warning: `yy_bp' might be used uninitialized in this
   > function
   
   I get no such warning.  Have you regenerated scan.c locally, and if so
   what version of flex did you use?
   
   			regards, tom lane
   
   

3. Re: scan.c:900: warning: ... postgresql-7.1

   Limin Liu <limin@pumpkinnet.com> — 2001-04-17T21:11:38Z

   > > scan.c: In function `base_yylex':
   > > scan.c:900: warning: `yy_cp' might be used uninitialized in this
   > > function
   > > scan.c:900: warning: `yy_bp' might be used uninitialized in this
   > > function
   >
   > I get no such warning.  Have you regenerated scan.c locally, and if so
   
   No, I didn't.  I un-tar the postgresql-7.1.tar.gz again and the scan.c
   line 900 shows no initialization.
   
   
   >
   > what version of flex did you use?
   >
   
   I have
   flex version 2.5.4
   
   Thanx
   
   
   --
   LM Liu
   
   
   
   

4. Re: scan.c:900: warning: ... postgresql-7.1

   Tom Lane <tgl@sss.pgh.pa.us> — 2001-04-17T21:23:44Z

   Limin Liu <limin@pumpkinnet.com> writes:
   >> I get no such warning.  Have you regenerated scan.c locally, and if so
   
   > No, I didn't.  I un-tar the postgresql-7.1.tar.gz again and the scan.c
   > line 900 shows no initialization.
   
   No, but it doesn't need to.  Both variables are assigned before first
   use, as even an extremely stupid compiler should be able to figure out.
   
   What compiler version are you using, and with what command-line switches?
   
   			regards, tom lane
   
   

5. Re: scan.c:900: warning: ... postgresql-7.1

   Limin Liu <limin@pumpkinnet.com> — 2001-04-17T22:03:36Z

   > No, but it doesn't need to.  Both variables are assigned before first
   > use, as even an extremely stupid compiler should be able to figure out.
   
   :-) I agree.
   
   >
   > What compiler version are you using, and with what command-line switches?
   >
   
   Here is what I have
   % gcc -v
   Reading specs from /usr/lib/gcc-lib/i386-redhat-linux/egcs-2.91.66/specs
   gcc version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release)
   
   I believe this compiler is not that old.
   
   After I un-tar the .gz, I run ./configure with prefix then gmake.  That's
   it.
   
   I first see warning in lex.Int_yy.c, than scan.c.    Here are those message,
   hope this can help.
   ======================
   gmake[3]: Entering directory `/tmp/tmp/postgresql-7.1/src/backend/bootstrap'
   
   gcc -O2 -Wall -Wmissing-prototypes -Wmissing-declarations -Wno-error
   -I../../../src/include   -c
   -o bootparse.o bootparse.c
   gcc -O2 -Wall -Wmissing-prototypes -Wmissing-declarations -Wno-error
   -I../../../src/include   -c
   -o bootscanner.o bootscanner.c
   lex.Int_yy.c: In function `Int_yylex':
   lex.Int_yy.c:689: warning: `Int_yy_cp' might be used uninitialized in this
   function
   lex.Int_yy.c:689: warning: `Int_yy_bp' might be used uninitialized in this
   function
        :
   gcc -O2 -Wall -Wmissing-prototypes -Wmissing-declarations -Wno-error
   -I../../../src/include   -c
   -o scan.o scan.c
   scan.c: In function `base_yylex':
   scan.c:900: warning: `yy_cp' might be used uninitialized in this function
   scan.c:900: warning: `yy_bp' might be used uninitialized in this function
        :
   pgc.c: In function `yylex':
   pgc.c:1243: warning: label `find_rule' defined but not used
   pgc.c:1178: warning: `yy_cp' might be used uninitialized in this function
   pgc.c:1178: warning: `yy_bp' might be used uninitialized in this function
   pgc.c: At top level:
   pgc.c:3091: warning: `yy_flex_realloc' defined but not used
   ======================
   configure does find flex and bison...
      :
   checking for flex... /usr/bin/flex
   checking whether ln -s works... yes
   checking for ld used by GCC... /usr/bin/ld
   checking if the linker (/usr/bin/ld) is GNU ld... yes
   checking for ranlib... ranlib
   checking for lorder... no
   checking for tar... /bin/tar
   checking for perl... perl
   checking for bison... bison -y
     :
   
   Everything looks fine to me and the previous copy (7.1 beta4) doesn't give
   me any warning message.  That's why I am curious what's going on...
   
   Thanks for your help
   
   --
   LM Liu
   
   
   
   

6. Re: scan.c:900: warning: ... postgresql-7.1

   Tom Lane <tgl@sss.pgh.pa.us> — 2001-04-17T22:50:39Z

   Limin Liu <limin@pumpkinnet.com> writes:
   >> What compiler version are you using, and with what command-line switches?
   
   > % gcc -v
   > Reading specs from /usr/lib/gcc-lib/i386-redhat-linux/egcs-2.91.66/specs
   > gcc version egcs-2.91.66 19990314/Linux (egcs-1.1.2 release)
   
   > I believe this compiler is not that old.
   
   Two years is a long time... I have been using gcc 2.95.2 until about two
   weeks ago, when I updated to 2.95.3.  Neither of them complain about
   this file, and I have not heard any reports of complaints from other
   people about it either.
   
   > Everything looks fine to me and the previous copy (7.1 beta4) doesn't give
   > me any warning message.  That's why I am curious what's going on...
   
   There were some irrelevant-looking changes in scan.l between beta4 and
   release, which perhaps could have triggered this glitch.
   
   Actually, what's even more interesting is that you get the same bogus
   warning messages from bootscanner.c.  I presume you weren't seeing that
   before, either?  bootscanner.l has not changed (except for a
   copyright-notice update) since 7.1beta1.  It's hard to draw any other
   conclusion than that something has gone wrong with your compiler
   installation.  Is it possible your disk dropped a bit, or something
   like that?
   
   			regards, tom lane
   
   

7. Re: scan.c:900: warning: ... postgresql-7.1

   Limin Liu <limin@pumpkinnet.com> — 2001-04-17T22:54:55Z

   > Two years is a long time... I have been using gcc 2.95.2 until about two
   > weeks ago, when I updated to 2.95.3.  Neither of them complain about
   > this file, and I have not heard any reports of complaints from other
   > people about it either.
   >
   
   I think I will update my compiler, since I have no clue some to answer your
   following question!  :-)
   
   >
   > installation.  Is it possible your disk dropped a bit, or something
   > like that?
   >
   
   Thank you very much
   
   --
   LM Liu
   
   
   
   

8. Need help with query

   Kevin Heflin <kheflin@shreve.net> — 2001-04-17T23:49:19Z

   Don't know if this would be the correct place to ask this. We have a 
   postgresql server running version 6.5.3
   It hosts a 100 or so different databases, and performs just great. I do 
   however have one query which takes about 20 seconds to complete. The 
   database only has 200-300 records. Don't know much about SQL other than 
   simple selects and inserts. I'm the one who created the query in 
   question, and even today when I look at it, I'm not sure I understand 
   it, but it gives us the results we want, it's just very slow.
   
   I'm sure someone with some SQL know-how could probably shorten this 
   query down some. But it's way over my head.
   
   Basically I've got a database with these tables:
   client (list of clients)
   project (each project belongs to a client)
   subproject (each subproject belongs to a project)
   status (each project has a status)
   users (each project and subproject are assigned to a user)
   timelog (each timelog record indicates time worked on a particular 
   subproject and who worked on it)
   
   A sample row would look like this:
   
   
   project_id|project_title|clientid|status|percentcomplete|duedate   
   |usersid|client_id|name       
   |users_id|users_name|status_id|status_name|hoursworked
   ----------+-------------+--------+------+---------------+----------+-------+
   ---------+-----------+--------+----------+---------+-----------+-----------
           143|project 1    |      79|     3|            
   100|04-06-2001|      8|       79|Client Name|       8|     Kevin|        
   3|status     |          1
   
   
   
   The query is below. Again if this is not the appropriate list for such a 
   question, my apologies.
   
   Kevin
   
   
   *******************
   SELECT
   p1.project_id, p1.project_title, p1.clientid, p1.status, 
   p1.percentcomplete, p1.duedate, p1.usersid,
   c2.client_id, c2.name,
   u3.users_id, u3.users_name,
   s4.status_id, s4.status_name,
   sum(case when t.jobid=s.subproject_id then t.timespent else 0::float4 
   end) as hoursworked
   
   FROM
   project p1, client c2, users u3, status s4, timelog t, subproject s
   
   WHERE
   (s4.status_id=2 OR s4.status_id=3) AND
   (c2.client_id=p1.clientid) AND (u3.users_id=p1.usersid) AND
   (s4.status_id=p1.status) AND (p1.project_id=s.projectid)
   
   GROUP BY
   p1.project_id, p1.project_title, p1.clientid, p1.status, 
   p1.percentcomplete, p1.duedate, p1.usersid,
   c2.client_id, c2.name, u3.users_id, u3.users_name, s4.status_id, 
   s4.status_name
   
   UNION SELECT
   p1.project_id, p1.project_title, p1.clientid, p1.status, 
   p1.percentcomplete, p1.duedate, p1.usersid,
   c2.client_id, c2.name,
   u3.users_id, u3.users_name,
   s4.status_id, s4.status_name,
   0 as hoursworked
   
   FROM
   project p1,client c2,users u3,status s4, subproject s
   
   WHERE
   (s4.status_id=2 OR s4.status_id=3) AND
   (c2.client_id=p1.clientid) AND (u3.users_id=p1.usersid) AND
   (s4.status_id=p1.status) AND
   
   NOT EXISTS
   (Select s.subproject_id from subproject s where 
   s.projectid=p1.project_id)
   
   GROUP BY
   p1.project_id, p1.project_title, p1.clientid, p1.status, 
   p1.percentcomplete, p1.duedate, p1.usersid,
   c2.client_id, c2.name, u3.users_id, u3.users_name, s4.status_id, 
   s4.status_name
   
   ORDER BY p1.project_title
   
   

9. Re: Need help with query

   Joel Burton <jburton@scw.org> — 2001-04-18T01:33:40Z

   On Tue, 17 Apr 2001, Kevin Heflin wrote:
   
   > Don't know if this would be the correct place to ask this. We have a 
   > postgresql server running version 6.5.3
   > It hosts a 100 or so different databases, and performs just great. I do 
   > however have one query which takes about 20 seconds to complete. The 
   > database only has 200-300 records. Don't know much about SQL other than 
   > simple selects and inserts. I'm the one who created the query in 
   > question, and even today when I look at it, I'm not sure I understand 
   > it, but it gives us the results we want, it's just very slow.
   > 
   > I'm sure someone with some SQL know-how could probably shorten this 
   > query down some. But it's way over my head.
   > 
   > Basically I've got a database with these tables:
   > client (list of clients)
   > project (each project belongs to a client)
   > subproject (each subproject belongs to a project)
   > status (each project has a status)
   > users (each project and subproject are assigned to a user)
   > timelog (each timelog record indicates time worked on a particular 
   > subproject and who worked on it)
   > 
   > A sample row would look like this:
   > 
   > 
   > project_id|project_title|clientid|status|percentcomplete|duedate   
   > |usersid|client_id|name       
   > |users_id|users_name|status_id|status_name|hoursworked
   > ----------+-------------+--------+------+---------------+----------+-------+
   > ---------+-----------+--------+----------+---------+-----------+-----------
   >         143|project 1    |      79|     3|            
   > 100|04-06-2001|      8|       79|Client Name|       8|     Kevin|        
   > 3|status     |          1
   
   Kevin --
   
   Looking quickly at your problem, I have one or two ideas, but need to play
   for a few minutes.
   
   Can you post a schema and a small amount of data?
   
   -- 
   Joel Burton   <jburton@scw.org>
   Director of Information Systems, Support Center of Washington