search

postgresql-interfaces / psqlodbc

Other
11 stars 10 forks source link

Notes todo and attempt to fix memory leak #18

Closed davecramer closed 3 months ago

davecramer commented 3 months ago

This may be too naive.

progmachine commented 3 months ago

This fix for memory leak is incorrect. It broke the meaning of CC_clear_col_info function - this function is clearing COL_INFO cache from ConnectionClass object, when it is obsolete, or connection is closing. Your fix will leave obsolete cache entries. To fix this leak correctly, we need to restore refcnt lifetime management. I have strong intent to do this, because these leaks causing fails with my project's unit tests. For now, i have another problem - many psqlodbc native tests are failing, saying that testtab1 does not exist... (

davecramer commented 3 months ago

This fix for memory leak is incorrect. It broke the meaning of CC_clear_col_info function - this function is clearing COL_INFO cache from ConnectionClass object, when it is obsolete, or connection is closing. Your fix will leave obsolete cache entries. To fix this leak correctly, we need to restore refcnt lifetime management. I have strong intent to do this, because these leaks causing fails with my project's unit tests. For now, i have another problem - many psqlodbc native tests are failing, saying that testtab1 does not exist... (

Well I need to figure out how to test this.

As for testtab1 why not use the existing tests ?

progmachine commented 3 months ago

As for testtab1 why not use the existing tests ?

I am talking about existing tests, located in test directory of psqlodbc project: regression.diffs.txt

false alarm... =)

davecramer commented 3 months ago

I guess you are running this on a mac or other *nix like machine.

The tests run fine on windows using regress.ps1