`CYGWIN=winsymlinks:nativestrict`, `ln -s target link` fails if target doesn't exist

Gene Pavlovsky gene.pavlovsky@gmail.com
Fri Apr 29 01:46:00 GMT 2016


The list of errors that might be returned by symlink(2) mention target
only 3 times:
       EFAULT target or linkpath points outside your accessible address space.
       ENAMETOOLONG
              target or linkpath was too long.
       ENOENT A directory component in linkpath does not exist or is a
dangling symbolic link, or target or linkpath is an empty string.

So my understanding is: target must be non-empty, not too long, and
don't point outside the accessible address space. Nothing about
existence.
--Gene


On 29 April 2016 at 03:05, Gene Pavlovsky <gene.pavlovsky@gmail.com> wrote:
> I don't know if POSIX standard has something to say about that, but
> here's a reference from GNU libc:
>
> `man 2 symlink`:
>>       A symbolic link (also known as a soft link) may point to an existing file or to a nonexistent one; the latter case is known as a dangling link.
>
> On 29 April 2016 at 02:06, Andrey Repin <anrdaemon@yandex.ru> wrote:
>> Greetings, Gene Pavlovsky!
>>
>>> I have an issue to report:
>>
>>> Introduction: On a UNIX system, `ln -s target link` creates a link
>>> regardless of target's existence.
>>> This is used in some scripts, e.g. Gentoo's `run-crons` (which I also
>>> use on Cygwin) uses a symlink pointing to the running process PID as
>>> lockfile.
>>> Issue: if `CYGWIN=winsymlinks:nativestrict` env var is set, running
>>> `ln -s target link` completely fails (even though running `mklink link
>>> target` in `cmd.exe` succeeds, same as `ln -s` does on UNIX). If
>>> `CYGWIN=winsymlinks:native`, a non-native link is created.
>>
>>> So, `nativestrict` might break some (admittedly unorthodox) scripts.
>>> With `native` these script work, but still a native link would be
>>> preferrable and it is possible to create, but a non-native link is
>>> created instead.
>>
>>> Bottom line, I think the native symlink creation code should be
>>> checked and a possibility should be added to create links to
>>> non-existent targets, rather than the current behavior of failing.
>>
>> This is actually an arguable behavior, even in Linux. I can imagine the
>> behavior is "undefined" in such a case.
>> But I'll leave final say to the more experienced members of the list.
>>
>>
>> --
>> With best regards,
>> Andrey Repin
>> Friday, April 29, 2016 01:55:21
>>
>> Sorry for my terrible english...
>>

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple



More information about the Cygwin mailing list