These have been inadvertently dropped when changing the test suite to
not require ext/interbase anymore, so we add them back.
We also change the required environment variable names to match the
usual PDO names. Particularly, we replace `PDO_FIREBIRD_TEST_HOSTNAME`
and `_DATABASE` with the more flexible `PDO_FIREBIRD_TEST_DSN`.
0x2d == 45, not 46. However, `CURLMOPT_PUSHFUNCTION` is actually
available as of cURL 7.44.0[1], while we make the respective PHP
constants available only as of cURL 7.46.0[2].
- This renames the environment variables to: PDO_FIREBIRD_TEST_XXX to be in line with other PDO test suites
- Adds an skipif.inc file that skips the tests if no database is set
- The test suite can now be run without PDO_FIREBIRD_TEST_HOSTNAME
$ set PDO_FIREBIRD_TEST_HOSTNAME=localhost
$ set PDO_FIREBIRD_TEST_DATABASE=C:\dev\php.fdb
$ nmake test TESTS=ext/pdo_firebird/tests
The only way I could get this to work on my local machine was with a DSN like:
- firebird:dbname=localhost:C:\php.fdb
Hence why the 'hostname' can be specified.
There is also a bit mess in regards to cleanup of this extension tests, but I really do not want to dwell down there, if someone dares to do so, then please be my guest.
Fix bug #77950 - Heap-buffer-overflow in _estrndup via exif_process_IFD_TAG
I do not completely understand what is going on there, but I am pretty
sure dir_entry <= offset_base if not a normal situation, so we better not
to rely on such dir_entry.
Fix #77821: Potential heap corruption in TSendMail()
`zend_string_tolower()` returns a copy (not a duplicate) of the given
string, if it is already in lower case. In this case we must not not
`zend_string_free()` both strings. The cleanest solution is to call
` zend_string_release()` on both strings, which properly handles the
refcount.
Peter Kokot [Mon, 29 Apr 2019 20:34:43 +0000 (22:34 +0200)]
Remove TSRM configuration header
TSRM configuration header file was once created by separate autoconf
build system for TSRM and is with the current code not directly needed
like this anymore.
Fix #77943: imageantialias($image, false); does not work
Firstly, we must not call `gdImageSetAntiAliased()` (which sets the
color to anti-alias), but rather modify the `gdImage.AA` flag.
Furthermore, we have to actually use the supplied boolean value.
We also make sure that we don't attempt to enable anti-aliasing for
palette images.
Fix #77943: imageantialias($image, false); does not work
Firstly, we must not call `gdImageSetAntiAliased()` (which sets the
color to anti-alias), but rather modify the `gdImage.AA` flag.
Furthermore, we have to actually use the supplied boolean value.
We also make sure that we don't attempt to enable anti-aliasing for
palette images.
The MS docs on `SQLColAttribute()`[1] state regarding the
`NumericAttributePtr` parameter:
| Please note that some drivers may only write the lower 32-bit or
| 16-bit of a buffer and leave the higher-order bit unchanged.
| Therefore, applications should initialize the value to 0 before
| calling this function.
Due to former restrictions of the libcurl API, curl multipart/formdata
file uploads supported only proper files. However, as of curl 7.56.0
the new `curl_mime_*()` API is available (and already supported by
PHP[1]), which allows us to support arbitrary *seekable* streams, which
is generally desirable, and particularly resolves issues with the
transparent Unicode and long part support on Windows (see bug #77711).
Note that older curl versions are still supported, but CURLFile is
still restricted to proper files in this case.
These tests are obviously meant to test successful and failing uri:
DSNs, but did not pass proper file:// URIs, so actually ended up
testing for invalid data source URIs twice. We fix this, and adjust
the expectations accordingly.
We also unfork the -win32 variant, since both test cases are almost
identical, and the expected error message may be either one.
Peter Kokot [Sat, 27 Apr 2019 15:21:56 +0000 (17:21 +0200)]
Simplify checking of *nix build tools
The buildmk.stamp file has been created by the *nix build checking step
to run the check step only once. Instead of poluting the project root
directory, the stamp file can be also omitted. Performance difference is
very minimal to not justify having the stamp check at all today anymore.
This patch integrates the buildcheck.sh to buildconf script directly.
Peter Kokot [Sun, 28 Apr 2019 03:15:47 +0000 (05:15 +0200)]
Remove unused TSRM/readdir.h header
This was once part of TSRM but then got refactored into the windows
implementation win32/readdir.h directly. Instead of including such files
directly code should use zend_virtual_cwd.h which is already part of the
php.h file.
MySQL/MariaDB version strings may have suffixes which may contain dots;
for instance, Debian stretch has 5.5.5-10.1.37-MariaDB-0+deb9u1 or
such. Therefore, we make the version extraction more liberal, and only
require that there are at least three parts separated by dot, and
ignore additional parts.
We also fix an erroneous test expectation, which would be triggered on
CI now, right away. This patch has been provided by petk@.
Peter Kokot [Sun, 28 Apr 2019 19:15:54 +0000 (21:15 +0200)]
Enhance the buildconf force option
The force option used to only remove the Autoconf cache files. Now it
also removes entire configure file so the configure file always gets
rebuilt when needs be done.
Additionally, the buildconf now also accepts the shorter version of the
force option:
buildconf -f
Peter Kokot [Thu, 25 Apr 2019 20:44:56 +0000 (22:44 +0200)]
Automatically remove aclocal.m4 if present
Commit 4e7064d173d2b5b22e159fcf52d22b10213b67b8 removed the usage of
`aclocal.m4`. When using Git repositories, many times cleaning of the
generated files is not done prior to running phpize or buildconf. For
example:
To not accidentally include `aclocal.m4` file in the generated configure
this enhances build system experience a bit more by removing aclocal.m4
file prior to start building configure file using phpize or buildconf.
Fix #77944: Wrong meta pdo_type for bigint on LLP64
When actually fetching the data, bigint (unsigned) column values are
returned as integers on LLP64 architectures, so their pdo_type has to
be PDO::PARAM_INT accordingly.
Add --enable-werror configure option and use it on Travis. It's not
possible to directly use CFLAGS, because it also affects configure
checks which often throw warnings.
We can't enable something similar for Windows builds at this time,
because they throw a lot more warnings.
We must not (try to) modify shared values, but rather have to use our
own copies, if unixified filenames are required on Windows. To avoid
excessive string duplication, we add checks whether the filenames are
already unixified (i.e. do not contain backslashes). To improve the
performance if we need to copy strings, we use do_alloca() and friends.
Besides generally being somewhat messy, the handling of unixified
filenames is still suboptimal performance-wise, but we leave this for a
future cleanup, and focus on fixing the issue at hand for now.
We also enable opcache.protect_memory for the AppVeyor CI.
projects / php / log