Thread: Source code cleanup
I'm getting a bunch of compiler warnings when I compile the latest psqlodbc sources from CVS. I spent some time cleaning those up. I also got access to a report of a Coverity scan over psqlodbc source tree. That found a bunch of additional issues, some of which were genuine (minor) bugs. I'm still in the process of going through the report, but I fixed some bugs already. I also wrote a small regression test suite to test with. It only tests a few of the ODBC functions at the moment, but it's better than nothing, and it can be easily extended. What do others use for testing psqlodbc? I pushed the result to a git repository at git@github.com:hlinnaka/psqlodbc.git. It contains the psqlodbc CVS history, as converted by git cvsimport, and my changes as commits on top of that. I can also send the changes as patches if that's preferred. Let me know how to proceed. - Heikki
Hi Heikki, (2013/01/09 3:25), Heikki Linnakangas wrote: > I'm getting a bunch of compiler warnings when I compile the latest > psqlodbc sources from CVS. I spent some time cleaning those up. > > I also got access to a report of a Coverity scan over psqlodbc source > tree. That found a bunch of additional issues, some of which were > genuine (minor) bugs. I'm still in the process of going through the > report, but I fixed some bugs already. Thanks. > > I also wrote a small regression test suite to test with. It only tests a > few of the ODBC functions at the moment, but it's better than nothing, > and it can be easily extended. What do others use for testing psqlodbc? > > I pushed the result to a git repository at > git@github.com:hlinnaka/psqlodbc.git. It contains the psqlodbc CVS > history, as converted by git cvsimport, and my changes as commits on top > of that. I would check the git repository. > I can also send the changes as patches if that's preferred. Let > me know how to proceed. > > - Heikki regards, Hiroshi Inoue
Hi Heikki, (2013/01/10 12:58), Inoue, Hiroshi wrote: > Hi Heikki, > > (2013/01/09 3:25), Heikki Linnakangas wrote: >> I'm getting a bunch of compiler warnings when I compile the latest >> psqlodbc sources from CVS. I spent some time cleaning those up. >> >> I also got access to a report of a Coverity scan over psqlodbc source >> tree. That found a bunch of additional issues, some of which were >> genuine (minor) bugs. I'm still in the process of going through the >> report, but I fixed some bugs already. > > Thanks. > >> >> I also wrote a small regression test suite to test with. It only tests a >> few of the ODBC functions at the moment, but it's better than nothing, >> and it can be easily extended. What do others use for testing psqlodbc? >> >> I pushed the result to a git repository at >> git@github.com:hlinnaka/psqlodbc.git. It contains the psqlodbc CVS >> history, as converted by git cvsimport, and my changes as commits on top >> of that. > > I would check the git repository. The 1st question I have is about the change commit 80b3eb35ab1428949f7ab8fbd567d5d49655dc02 Author: Heikki Linnakangas <heikki.linnakangas@iki.fi> Date: Sat Jan 5 00:35:39 2013 +0200 Silence Coverity warnings about passing negative param to SQLExecDirect. It doesn't know about the psqlodbc extension to that function that passing SQL_NTS means it's a C-style null-terminated string. The type of the 3rd parameter of SQLExecDirect() is SQLINTEGER. What kind od warnings does Coverity scan show? SQL_NTS is generally used in ODBC and I could hardly remember the examples which use parameters other than SQL_NTS. In addition the change doesn't seem correct to me. The 3rd parameter of the following SQLExecDirect call is the string length of a Statement Handle. ret = SQLExecDirect(stmt, (SQLCHAR *) "select gid from pg_prepared_xacts", strlen(stmt)); regards, Hiroshi Inoue
(cc'ing pgsql-odbc mailing list yet again, I misspelled it earlier..) On 22.02.2013 00:39, Hiroshi Inoue wrote: > HI Heikki, > > (2013/02/21 17:55), Heikki Linnakangas wrote: >> (cc'ing psql-odbc mailing list again, got dropped by accident) >> >> Hi Hiroshi, >> >> Any news on this? Would you like me to prepare a new set of patches with >> those things fixed? > > Sorry for the late reply. > I'm happy if you make new set of patches. Ok, I fixed the couple of bugs you mentioned, rebased, and pushed to the git repository again. See master branch at git@github.com:hlinnaka/psqlodbc.git. Link to the github web interface: https://github.com/hlinnaka/psqlodbc. I separated the regression test suite I wrote to a separate branch, master-with-testcases, for easier review. There's also a third branch in the repository, "upstream", which is a mirror of the current CVS tip. I kept the change that turns the switch- to if-statement in getPrecisionPart() for now (https://github.com/hlinnaka/psqlodbc/commit/1dd3bfb6e6a7f5d1bef49103593178ccc2caf129). You suggested adding a default-case to that instead, which is fine with me as well so feel free to change it at commit, although I find it slightly more readable with "if". - Heikki