On Windows data can be lost in buffers in case of abnormal program
termination, especially in process chains as seen due to flaky tests.
Therefore flushing all buffers manually should avoid this data loss.
In the curl tool we play the safe game by only flushing write buffers,
but in the testsuite where we manage all buffers, we flush everything.
This should drastically reduce Windows CI and testsuite flakiness.
Reviewed-by: Daniel Stenberg
Supersedes #7833 and #6064Closes#8516
In March 2010 (commit 4259d2df7d) we removed the embedded 'ares'
directory from the curl source tree but we have since supported
especially detecting and using that build directory. The time has come
to remove that kludge and ask users to specify the c-ares dir correctly
with --enable-ares.
Closes#8397
There has been no TPF related changes done since September 2010 (commit
7e1a45e224) and since this is a platform that is relatively different
than many others (== needs attention), I draw the conclusion that this
build is broken since a long time.
Closes#8378
The callbacks were partially documented to support this. Now the
behavior is documented and returning error from either of these
callbacks will effectively kill all currently ongoing transfers.
Added test 530 to verify
Reported-by: Marcelo Juchem
Fixes#8083Closes#8089
Previously, the return code CURLUE_MALFORMED_INPUT was used for almost
30 different URL format violations. This made it hard for users to
understand why a particular URL was not acceptable. Since the API cannot
point out a specific position within the URL for the problem, this now
instead introduces a number of additional and more fine-grained error
codes to allow the API to return more exactly in what "part" or section
of the URL a problem was detected.
Also bug-fixes curl_url_get() with CURLUPART_ZONEID, which previously
returned CURLUE_OK even if no zoneid existed.
Test cases in 1560 have been adjusted and extended. Tests 1538 and 1559
have been updated.
Updated libcurl-errors.3 and curl_url_strerror() accordingly.
Closes#8049
file URLs that are 6 bytes or shorter are not complete. Return
CURLUE_MALFORMED_INPUT for those. Extended test 1560 to verify.
Triggered by #8041Closes#8042
In order to check the actual code better, checksrc now ignores
everything that look like preprocessor instructions. It also means
that code in macros are now longer checked.
Note that some rules then still don't need to be followed when code is
exactly below a cpp instruction.
Removes two checksrc exceptions we needed previously because of
preprocessor lines being checked.
Reported-by: Marcel Raad
Fixes#7863Closes#7864
The host name is stored decoded and can be encoded when used to extract
the full URL. By default when extracting the URL, the host name will not
be URL encoded to work as similar as possible as before. When not URL
encoding the host name, the '%' character will however still be encoded.
Getting the URL with the CURLU_URLENCODE flag set will percent encode
the host name part.
As a bonus, setting the host name part with curl_url_set() no longer
accepts a name that contains space, CR or LF.
Test 1560 has been extended to verify percent encodings.
Reported-by: Noam Moshe
Reported-by: Sharon Brizinov
Reported-by: Raul Onitza-Klugman
Reported-by: Kirill Efimov
Fixes#7830Closes#7834
Triggered before a request is made but after a connection is set up
Changes:
- callback: Update docs and callback for pre-request callback
- Add documentation for CURLOPT_PREREQDATA and CURLOPT_PREREQFUNCTION,
- Add redirect test and callback failure test
- Note that the function may be called multiple times on a redirection
- Disable new 2086 test due to Windows weirdness
Closes#7477
- file://host.name/path/file.txt is a valid UNC path
\\host.name\path\files.txt to a non-local file transformed into URI
(RFC 8089 Appendix E.3)
- UNC paths on other OSs must be smb: URLs
Closes#7366
Add curl_url_strerror() to convert CURLUcode into readable string and
facilitate easier troubleshooting in programs using URL API.
Extend CURLUcode with CURLU_LAST for iteration in unit tests.
Update man pages with a mention of new function.
Update example code and tests with new functionality where it fits.
Closes#7605
When setting a blank expire string, meaning unlimited, curl would pass
TIME_T_MAX to getime_r() when creating the output, while on 64 bit
systems such a large value cannot be convetered to a tm struct making
curl to exit the loop with an error instead. It can't be converted
because the year it would represent doesn't fit in the 'int tm_year'
field!
Starting now, unlimited expiry is instead handled differently by using a
human readable expiry date spelled out as "unlimited" instead of trying
to use a distant actual date.
Test 1660 and 1915 have been updated to help verify this change.
Reported-by: Jonathan Cardoso
Fixes#7720Closes#7721
- Use our wait_ms() instead of sleep() since Windows doesn't have the
latter.
- Use a separate variable to keep track of whether the pthread_t thread
id is valid.
On Windows pthread_t is not an integer type. pthread offers no macro for
invalid pthread_t thread id, so validity is kept track of separately.
Closes https://github.com/curl/curl/pull/7527
The length of 'long' in a 32-bit system is 32 bits, which cannot be used
to save timestamps after 2038. Most operating systems have extended
time_t to 64 bits.
Remove the castings to long.
Closes#7466
They were never officially allowed and slipped in only due to sloppy
parsing. Spaces (ascii 32) should be correctly encoded (to %20) before
being part of a URL.
The new flag bit CURLU_ALLOW_SPACE when a full URL is set, makes libcurl
allow spaces.
Updated test 1560 to verify.
Closes#7073
Warning: this will make existing curl command lines that use metalink to
stop working.
Reasons for removal:
1. We've found several security problems and issues involving the
metalink support in curl. The issues are not detailed here. When
working on those, it become apparent to the team that several of the
problems are due to the system design, metalink library API and what
the metalink RFC says. They are very hard to fix on the curl side
only.
2. The metalink usage with curl was only very briefly documented and was
not following the "normal" curl usage pattern in several ways, making
it surprising and non-intuitive which could lead to further security
issues.
3. The metalink library was last updated 6 years ago and wasn't so
active the years before that either. An unmaintained library means
there's a security problem waiting to happen. This is probably reason
enough.
4. Metalink requires an XML parsing library, which is complex code (even
the smaller alternatives) and to this day often gets security
updates.
5. Metalink is not a widely used curl feature. In the 2020 curl user
survey, only 1.4% of the responders said that they'd are using it. In
2021 that number was 1.2%. Searching the web also show very few
traces of it being used, even with other tools.
6. The torrent format and associated technology clearly won for
downloading large files from multiple sources in parallel.
Cloes #7176
