Commits on Sep 18, 2008
  1. Tagged release of pgTAP 0.10.

    theory committed Sep 18, 2008
  2. Timestamped for 0.10 release.

    theory committed Sep 18, 2008
  3. Typo.

    theory committed Sep 18, 2008
  4. markdown errors.

    theory committed Sep 18, 2008
  5. * Incremented the version number to 0.10.

    theory committed Sep 18, 2008
     * Edited `Changes`.
Commits on Sep 17, 2008
  1. Versions tested.

    theory committed Sep 17, 2008
  2. * No longer installing the test scripts.

    theory committed Sep 17, 2008
     * Updated the docs a bit more. Now have to specify `$TAPSCHEMA` for the `installcheck` and `test` targets, too.
  3. Keyword.

    theory committed Sep 17, 2008
  4. Updated docs to reflect that `throws_ok()` and `lives_ok()` are not s…

    theory committed Sep 17, 2008
    …upported under 8.0.
  5. * Updated installation docs.

    theory committed Sep 17, 2008
     * Updated the uninstall script.
  6. * Added the `--pset` option to `pg_prove`, and made use of it in the…

    theory committed Sep 17, 2008
    … `test` target so that it is always properly toggled.
     * Removed the `--pset null=[NULL]` call in `pg_prove`, since users can now use `--pset` to do it themselves.
  7. Use the command-line parameter.

    theory committed Sep 17, 2008
  8. Didn't mean to delete that.

    theory committed Sep 17, 2008
  9. Ported to PostgreSQL 8.0. I did most of the hard work the other day, …

    theory committed Sep 17, 2008
    …and realized that I could apply it all using a patch. So that's what I've done. Details:
     * Added an `ORDER BY` clause to the `SELECT` statement that fetches the list of foreign key column combinations in `col_is_fk()`. This ensures that the output is always the same, which is at least important for my test.
     * Updated `Makefile` to replace all instances of `E''` when using 8.0.
     * Updated `Makefile` to apply a patch to get things working on 8.0.
     * Changed a test in `sql/` so that there is a space before a `E'`. This allows the substitution to work without needlessly killing off strings that end in "E'".
     * Updated `Makefile` so that the list of test scripts is generated. It also strips out the `throwtap` test in 8.0, since `throws_ok()` and `lives_ok()` are not supported in 8.0.
  10. Compatibility in tests for the TAPSCHEMA. Just be sure to use "public…

    theory committed Sep 17, 2008
    …" when we are going to be testing "public".
  11. Restore comaptibility with 8.3.

    theory committed Sep 17, 2008
  12. Ported pgTAP to PostgreSQL 8.2. Mostly this involved adding the appro…

    theory committed Sep 17, 2008
    …priate casts. Here's how I did it:
     * Modified `Makefile` to concatenate extra compatibility files into `pgtap.sql` and `uninstall_pgtap.sql` depending on version numbers. It's ugly, but it works well enough. The files live in `compat` and are named for the appropriate version. Right now that means that it adds the contents of extra files for 8.2. I'll be adding more for 8.1 later, if necessary. And more for 8.0 even later. Maybe.
     * Added some temporary functions to `sql/cmp_ok.sql`. These allow the tests to properly run in 8.2 and older, where those functions are not available. They're not required for the test suite itself, though, just for these particular tests. 
     * Added compatibility files for PostgreSQL 8.2. It includes a few casts and one operator.
     * While working on this stuff, added the `skip()` function. I need to write some tests for it, though.
     * Eliminated unnecessary casts to `::text` in calls to `quote_literal()`. I had them in one place but not another.
  13. Reverted some of the changes for 8.0. I'm going to go back to 8.2 fir…

    theory committed Sep 17, 2008
    …st, then 8.1, and then 8.0 if I really feel like it.
Commits on Sep 16, 2008
  1. Started making changes to supports PostgreSQL 8.0:

    theory committed Sep 16, 2008
     * Older versions of PostgreSQL cannot cast `name[]` to text. So we pass such variables through `array_to_string()`, instead. I think that the output is probably more appropriate, anyway.
     * Replaced all but one usage of `regexp_replace()`, since it's not available in older versions of PostgreSQL. That last one will be replaced before long, too.
     * Switched to using `FOR record IN EXECUTE` to fetch a value. It's an ugly hack, but older versions of PostgreSQL require it. I'll probably have to do the same thing in a few other places.
     * Booleans don't case to text very well in 8.0. So use a CASE statement, instead. Thanks to David Westerbrook for the patch.
     * Fixed the raising of an exception so that its variable argument is not dynamically generated, which 8.0 doesn't seem to like.
Commits on Sep 15, 2008
  1. Let's be consistent, shall we?

    theory committed Sep 15, 2008
  2. Better example.

    theory committed Sep 15, 2008
  3. * Updated documentation on the diagnostics from `col_is_fk()`.

    theory committed Sep 15, 2008
     * Tweaked said diagnostic message to be a bit more accurate.
  4. Switched schema testing functions to use `name` parameters instead of…

    theory committed Sep 15, 2008
    … `text` where appropriate.
  5. Added tests for `col_is_fk()` when there are no FK columns in the tar…

    theory committed Sep 15, 2008
    …get table. Tweaked the diagnostics there, too.
  6. Fixed `col_is_fk()` to use the new view. Also improved the diagnostic…

    theory committed Sep 15, 2008
    …s on failure to just list the available FK columns, if there are any.
  7. Added a view from newsysviews in order to make querying for foreign k…

    theory committed Sep 15, 2008
    …eys easier. And it's a lot better now, with better support for multiple FKs and the like. Still need to update the `col_is_fk()` functions to use it. That's up next.
Commits on Sep 12, 2008
  1. Fixed function for fetching PKs for FKs. Still have new failing tests…

    theory committed Sep 12, 2008
    …, though, as I need to correct for the opposite.
Commits on Sep 7, 2008
  1. * Changed output of extra tests so user does not have to do any extr…

    theory committed Sep 7, 2008
    …a math.
     * Added some functions to the To Do list, so I remember to implement them.
Commits on Sep 5, 2008
  1. Updated uninstall script and fixed formatting in main script to make …

    theory committed Sep 5, 2008
    …uninstall script generation easier in the future.