Re: Branch 1.14? - Mailing list pgadmin-hackers
| From | Jasmin Dizdarevic |
|---|---|
| Subject | Re: Branch 1.14? |
| Date | |
| Msg-id | BANLkTikEshJMBfPodYUi6GRUwaghDQzQug@mail.gmail.com Whole thread Raw |
| In response to | Re: Branch 1.14? (Guillaume Lelarge <guillaume@lelarge.info>) |
| Responses |
Re: Branch 1.14?
(Guillaume Lelarge <guillaume@lelarge.info>)
|
| List | pgadmin-hackers |
currently there are 10 versions of pgadmin for download (http://pgadmin.org/download/windows.php) available, which works with older versions. if someone likes to manage an older version he can download an older client.
there are no tests of current releases against older versions of postgresql, so why we should keep the code?
the code would become more clear and readable, because we would not have to check things which are a must. take these lines of code:
if (collection->GetConnection()->BackendMinimumVersion(7, 4))
{
proname = wxT("indnatts, ");
if (collection->GetConnection()->BackendMinimumVersion(7, 5))
{
proname += wxT("cls.reltablespace AS spcoid, spcname, ");
projoin = wxT(" LEFT OUTER JOIN pg_tablespace ta on ta.oid=cls.reltablespace\n");
}
}
else
{
proname = wxT("proname, pn.nspname as pronspname, proargtypes, ");
projoin = wxT(" LEFT OUTER JOIN pg_proc pr ON pr.oid=indproc\n")
wxT(" LEFT OUTER JOIN pg_namespace pn ON pn.oid=pr.pronamespace\n");
}
query = wxT("SELECT DISTINCT ON(cls.relname) cls.oid, cls.relname as idxname, indrelid, indkey, indisclustered, indisunique, indisprimary, n.nspname,\n")
wxT(" ") + proname + wxT("tab.relname as tabname, indclass, con.oid AS conoid, CASE contype WHEN 'p' THEN desp.description WHEN 'u' THEN desp.description WHEN 'x' THEN desp.description ELSE des.description END AS description,\n")
wxT(" pg_get_expr(indpred, indrelid") + collection->GetDatabase()->GetPrettyOption() + wxT(") as indconstraint, contype, condeferrable, condeferred, amname\n");
if (collection->GetConnection()->BackendMinimumVersion(8, 2))
query += wxT(", substring(array_to_string(cls.reloptions, ',') from 'fillfactor=([0-9]*)') AS fillfactor \n");
query += wxT(" FROM pg_index idx\n")
wxT(" JOIN pg_class cls ON cls.oid=indexrelid\n")
wxT(" JOIN pg_class tab ON tab.oid=indrelid\n")
+ projoin +
wxT(" JOIN pg_namespace n ON n.oid=tab.relnamespace\n")
wxT(" JOIN pg_am am ON am.oid=cls.relam\n")
wxT(" LEFT JOIN pg_depend dep ON (dep.classid = cls.tableoid AND dep.objid = cls.oid AND dep.refobjsubid = '0' AND dep.refclassid=(SELECT oid FROM pg_class WHERE relname='pg_constraint') AND dep.deptype='i')\n")
wxT(" LEFT OUTER JOIN pg_constraint con ON (con.tableoid = dep.refclassid AND con.oid = dep.refobjid)\n")
wxT(" LEFT OUTER JOIN pg_description des ON des.objoid=cls.oid\n")
wxT(" LEFT OUTER JOIN pg_description desp ON (desp.objoid=con.oid AND desp.objsubid = 0)\n")
wxT(" WHERE indrelid = ") + collection->GetOidStr()
+ restriction + wxT("\n")
wxT(" ORDER BY cls.relname");
with a "supporting 8.2 and above" - policy, this would be just ONE statement and it's done.
2011/6/15 Guillaume Lelarge <guillaume@lelarge.info>
On Tue, 2011-06-14 at 23:41 +0100, Dave Page wrote:I don't see why we should remove old code that still works. That we say
> We don't tend to remove the old code as that can lead to painful
> amounts of refactoring to make things look right afterwards. If you
> feel so inclined though, go ahead and do some cleanup.
>
we don't "officially" support older versions than 8.2, that's one thing
I would understand. Working on dropping code that should work to make
sure no one can use it with older versions is something I can't
understand.
So, -1 for dropping any code, unless we have proof it contains a bug we
don't want to fix because it's on some old unsupported PostgreSQL
release.
(and when I say "-1", I really want to say "-1000000").
--
pgadmin-hackers by date: