| Age | Commit message (Collapse) | Author |
|---|
|
The addition of support for PostgreSQL 9.6 and above neglected to change the `Postgres94` class to inherit from `Postgres95` instead of the latest `Postgres` class. This change restores the chain of inheritance, so that `Postgres94` and classes below see overridden methods from `Postgres95` and above.
|
|
Based on code and suggestions from @cathysax, ultimately I used the internal
pg_has_role function to test whether a user has ownership rights based on
group membership. I actually check for 'USAGE' rights, since that implies the
role has rights without need to `set role`, which users wouldn't be able to do
with a normal PPA login. Loosely tested back to 9.5.
This fixes https://github.com/phppgadmin/phppgadmin/issues/102
|
|
For version 14, we point to the /devel/ docs, since that version is currently
in dev. I also bumped the fall through to point to v14, arguably we could
instead point to /current/, but any current version will work, so I think
devel is more likely.
|
|
|
|
This is mostly just refactoring Nirgals patch to use a standard
capability function. I also short-circuited default_with_oid check since it
will always show false in >=PG12, so this saves a round trip. This also adds
a comment in the config file that show oids is no longer supported. This
could lead to a slight difference in running instalations, but since it
does not affect any code/behavior, not going to bump the config version.
|
|
Notes:
- "CREATE TABLE WITHOUT OIDS" continues to work, for now
- "SHOW default_with_oids" continues to work, for now
|
|
|
|
|
|
|
|
|
|
Whilst fixing the sequence group viewing bug, noticed that sequence properties
in general would throw an error if you did not have appropriate permissions.
While this is similar to permission denied errors when looking directly at
a sequence, there is other information to be exposed which is accesible
through pg_sequence and pg_sequences. The change implements a method for
determining if a user can access the sequence directly (ie. select * from seq)
and if so, include that information in the output. If not, we null those
columns.
This also fixes cases where last_value information is not visible due to a
sequence not haveing been "read" yet. ie. after sequence creation, or anytime
a sequence is altered, and nextval/setval hasn't been called.
This problem seems to go back to at least postgres 9, so I've mainlined the fix.
It might break on older unsupported versions of postgres; if someone wants to
fix that well... patches welcome.
|
|
While I have implemented this fix against Postgres 12, as far as I can tell,
this bug existed since 7.x, so I'm adding it to all versions back to 7.4
(when we did the class re-org). This could probably use more testing on older
versions though.
Thanks to Carlos Martinez for the bug report and reproducible test case!
|
|
Fix based on patches from @nirgal and @LuckySB
Reported at:
https://sourceforge.net/p/phppgadmin/bugs/448/
https://bugs.debian.org/762378
https://github.com/phppgadmin/phppgadmin/pull/27
https://github.com/phppgadmin/phppgadmin/pull/30
|
|
chmod -x all php files without a #!
chmod +x those with one.
|
|
window functions, breaking functions, aggregates, and object look-up within
PPA. This commit fixes all three of these areas. In theory we should probably
add some display bits to help users differentiate between these different
types in the UI, but this at least makes the system work again.
Thanks to @decodeJarvis who initially reported the problem and provided a
proposed fix (see PR#24 on the xzilla repo). Since we needed a more
comprehensive fix, I didn't use the code, but his patch was useful for getting
started.
|
|
|
|
This commit is focused strictly on the class/constructor changes. This has
primarily been tested on php 7.1 and postgres 11, with all tests passing though
there are some spurious warnings; I have fixes for them but will add them in
seperate commits.
Note that the code here is my own, but I did look at patches from
@gabrielhomsi and @w1ldzer0 for some sanity checking. Any bugs or missing
items are on me.
|
|
supported versions.
|
|
|
|
|
|
fields. Inspired by patch from Felipe Figueroa <amenadiel@gmail.com>
|
|
|
|
|
|
Based on work by Bert Thomas <bthomas@brothom.nl>
|
|
See discussion on bug #441:
http://sourceforge.net/p/phppgadmin/bugs/441/
|
|
|
|
Also add in the waiting option, making it work back to at least 8.1
|
|
|
|
Fix bugs introduced in commit d571ecae7b by ioguix on 2012-09-21.
I can't tell why these changes crept in along with the rest of
the fixes in that commit. It looks like a testing version was
committed.
|
|
|
|
bytea columns.
Based on the POC patch from ioguix.
This fixes SF Bug # 3243916.
|
|
These queries need to be classoid qualified, so they work when there
are multiple objects with the same oid but of different classes.
(Which can always happen, but is a lot more likely to happen after
a cluster has been pg_upgrade'd)
|
|
Sourceforge Bug ID: 3579602.
|
|
|
|
Autovacuum under 8.1 doesn't has vacuum_freeze_min_age and
autovacuum_freeze_max_age.
|
|
|
|
Fix a bug INSERT SQL that was using 'WHERE' (?!)
|
|
Field pg_catalog.pg_stat_activity.procpid is renamed as pid in 9.2
|
|
Patch by Leonardo Sapiras, remerged by ioguix.
* remove old plugin architecture
* remove plugin Slony
|
|
|
|
passes tests for 90,91,92(ish)
|
|
|
|
When renaming a tablespace, it was renamed first, then a comment was set
on its old name in the same xact, leading to an error and rollback.
|
|
Translations, db connexions and html pages are now all in UTF-8. In
consequence:
* we remove everything about recode stuffs
* recode all lang/ translation files to UTF-8
* force UTF-8 in htmlentities
|
|
|
|
|
|
superUser status of the current role
Fixed a bug on the way as in previous version it was checking if function
pg_parameter_status exists and used it directly, hence testing only the
currently used role.
pg_parameter_status is available in PHP 5.*
|
|
|
|
|
|
database access classes
|
