[CALL FOR TESTING] Cygwin-3.6.0

Wed Feb 19 00:20:59 GMT 2025

Something is very wrong with bash. If I redirect the output of xcopy
or icacls into a bash variable, like stdout="$(icacls "$(cygpath -w
"$PWD")")", then sometimes I get only a letter in the middle. Very
strange
Lionel
On 2025年2月18日 at 10:37, Corinna Vinschen via Cygwin
<cygwin@cygwin.com> wrote:
>>> [I sent this on 2024年02月14日 to the cygwin-announce mailing list,
> but forgot to CC the cygwin mailing list. Doing it now...]
>>> Hi folks,
>> we're planning the next major release in the next few (hopefully
> not more than two) weeks.
>> We're looking forward to more people giving the latest test release
> of Cygwin a try!
>> The most recent test release, which you can easily install from Cygwin
> Setup, is cygwin-3.6.0-0.374.g4dd859d01c22. Of course, as time goes by,
> there may be more 3.6.0 test release prior to the official release.
>> If you're unsure how to install a test release, please have a look
> at https://cygwin.com/faq.html#faq.setup.testrels
>> At this point, we're most interested if the test release introduces
> regressions compared to 3.5.7!
>> There's a whole bunch of new features and a few behavioral changes
> compared to 3.5, but it's not so much testing the new functionality
> we're after, but to test if everything still runs as in 3.5.7 or
> even... better!
>> If you see a change in the test release which breaks something for
> you running fine in 3.5.7, please send a mail, ideally with a short
> and simple testcase, to the cygwin@cygwin.com mailing list.
>> Below you'll find a list of changes compared to the 3.5 series.
>>> Thanks a lot,
> Corinna
>>> What's new:
> -----------
>> - New API call: getlocalename_l.
>> - New API calls: fdclosedir, posix_getdents.
>> - New API call: setproctitle.
>> - New API call: timespec_get.
>> - New API calls: tcgetwinsize, tcsetwinsize.
>> - New POSIX-defined entry points posix_spawn_file_actions_addchdir and
> posix_spawn_file_actions_addfchdir. These are the same as the already
> exported posix_spawn_file_actions_addchdir_np and
> posix_spawn_file_actions_addfchdir_np.
>> - Add FS_IOC_[GS]ETFLAGS FS_PINNED_FL and FS_UNPINNED_FL flags to handle
> Windows attributes FILE_ATTRIBUTE_PINNED and FILE_ATTRIBUTE_UNPINNED.
> Add matching 'p' and 'u' mode bits in chattr(1) and lsattr(1).
>> - New libaio.a provided for projects checking for POSIX aio support
> by looking for this library at configure time.
>> - cygpath -r option allows to generate all Windows paths with root-local
> path prefix \\?\.
>>> What changed:
> -------------
>> - ps -f now prints the commandline rather than the full path to the
> executable.
>> - Drop support for NT4 and Samba < 3.0.22.
>> - Now that SMBv1 is ultimately deprecated and not installed by default
> on latest Windows versions, use Network Discovery (i. e. WSD, "Web
> Service Discovery") for enumerating network servers in //, just like
> Windows Explorer.
>> - If "server" is given as FQDN, and if "server" is an NFS server,
> ls //server now also enumerates NFS shares. If "server" is given
> as a flat name, only SMB shares are enumerated.
>> - Expose //tsclient (Microsoft Terminal Services) shares as well as
> //wsl$ (Plan 9 Network Provider) shares, i. e., WSL installation
> root dirs.
>> - Redesign pipe handling to minimize toggling blocking mode.
> The query_hdl stuff is no longer needed in new implementation.
>> - Now using AVX/AVX2/AVX-512 instructions in signal handler does not
> break their context.
>> - nice(2), setpriority(2) and sched_setparam(2) now fail with EACCES
> or EPERM if Windows would silently set a lower priority
> (HIGH_PRIORITY_CLASS instead of REALTIME_PRIORITY_CLASS) due to
> missing administrator privileges.
>> - nice(2) now returns the new nice value instead of 0 on success
> and sets errno to EPERM instead of EACCES on failure. This confirms
> to POSIX and Linux (glibc >= 2.2.4) behavior.
>> - sched_setscheduler(2) now emulates changes between SCHED_OTHER,
> SCHED_BATCH, SCHED_IDLE, SCHED_FIFO and SCHED_RR. If SCHED_OTHER or
> SCHED_BATCH is selected, the Windows priority is set according to the
> nice value where SCHED_BATCH sets a one step lower priority. If
> SCHED_IDLE is selected, the nice value is preserved and the Windows
> priority is set to IDLE_PRIORITY_CLASS. If SCHED_FIFO or SCHED_RR is
> selected, the nice value is preserved and the Windows priority is set
> according to the realtime priority.
> If the SCHED_RESET_ON_FORK flag is set, SCHED_FIFO and SCHED_RR are
> reset to SCHED_OTHER and negative nice values are reset to zero in
> each child process created with fork(2).
> Note: Windows does not offer alternative scheduling policies so
> this could only emulate API behavior.
>> - If SCHED_FIFO or SCHED_RR is selected, the /proc/<PID>/stat field
> '(18) priority' is now set to the negated sched_policy minus one.
> If SCHED_IDLE is selected, this field is set to 39. The '(19) nice'
> field is now set to the originally requested nice value.
>> - Raise maximum pid from 65536 to 4194304 to account for scenarios
> with lots of CPUs and lots of tasks.
> Addresses: https://cygwin.com/pipermail/cygwin/2024-December/256927.html
>> - Allow mmap with MAP_FIXED to succeed on an address range contained
> in the chunk of an existing anonymous mapping, provided the
> MAP_SHARED/MAP_PRIVATE flags agree and MAP_NORESERVE is not set for
> either mapping.
> Addresses: https://cygwin.com/pipermail/cygwin/2024-December/256901.html
>> - Fix a long-standing hang issue when running on ARM64 under emulation.
> This was due to a thread being terminated while the emulation was
> holding an internal lock.
> Addresses: https://cygwin.com/pipermail/cygwin-developers/2024-May/012694.html
>> - Add a host machine tag to uname(2)'s sysname field. This echoes what
> used to be done with -WOW64 (when that was supported), but now with
> -ARM64 when running on an ARM64 host under emulation. The Cygwin DLL's
> own architecture continues to be reported in the machine field.
>> - Escape special characters in /proc/<PID>/mount*. This allows the
> contents to be parsed consistently, and matches what is done on Linux.
> Addresses: https://cygwin.com/pipermail/cygwin/2024-June/256082.html
>> - Expose all Windows volume mount points via getmntent(3). This also
> exposes them via /proc/<PID>/mount*. A change in behavior from
> previous Cygwin versions is that volumes whose root is mounted
> explicitly in Cygwin will now also show up as mounted under the
> cygdrive prefix, whereas before that entry would have been suppressed.
> Addresses: https://cygwin.com/pipermail/cygwin/2024-June/256081.html
>>> Fixes:
> ------
>> - Fix an off-by-one bug in lseek(2)'s SEEK_DATA/SEEK_HOLE handling on
> filesystem-compressed files, potentially triggering a hang in cp(1).
> Addresses: https://sourceware.org/pipermail/cygwin/2025-January/257082.html
> https://cygwin.com/pipermail/cygwin/2025-February/257326.html
> --
> *** CYGWIN-ANNOUNCE UNSUBSCRIBE INFO ***
>> The easiest way to unsubscribe is to visit <https://cygwin.com/mailman/options/cygwin-announce>, and click 'Unsubscribe'.
>> If you need more information on unsubscribing, start reading here: <https://sourceware.org/lists.html#unsubscribe>.
>> --
> Problem reports: https://cygwin.com/problems.html
> FAQ: https://cygwin.com/faq/
> Documentation: https://cygwin.com/docs.html
> Unsubscribe info: https://cygwin.com/ml/#unsubscribe-simple

-- 
Lionel