[Tcl-bugs] [ tcl-Bugs-2388866 ] [file tempfile] can only create file in native fs

[SourceForge.net <no...@so...>]

Bugs item #2388866, was opened at 2008-12-04 14:30
Message generated for change (Settings changed) made by dkf
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=110894&aid=2388866&group_id=10894

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: 16. Commands A-H
Group: development: 8.6a4
Status: Open
Resolution: None
>Priority: 4
Private: No
Submitted By: Don Porter (dgp)
Assigned to: Donal K. Fellows (dkf)
Summary: [file tempfile] can only create file in native fs

Initial Comment:

[file tempfile var template] is supposed
to use the template as a pattern on what
pathname to give the temporary file.  If
the directory part of the template points
into a mounted vfs, though, this cannot
work, since the deep guts of [file tempfile]
are written in terms of native syscalls
such as mkstemps() which know nothing about
Tcl's collection of virtual filesystems.

[fileutil::tempfile] has other design
troubles, but it does not suffer from this
limitation.  I don't see why we should
accept it from [file tempfile].



----------------------------------------------------------------------

>Comment By: Donal K. Fellows (dkf)
Date: 2008-12-06 20:15

Message:
Forced to use the native FS for now; attempting to put the file in a
virtual FS makes that part of the template get dropped instead. This means
that the temporary is available for use with OS APIs and external programs.

----------------------------------------------------------------------

Comment By: Jeffrey Hobbs (hobbs)
Date: 2008-12-04 18:23

Message:
I would say this is a limitation of the implementation that can be fixed
for 8.6.0.  All you need to do is check if template points to a non-native
area and fallback to a virtual file.  I could see this potentially having
issues with certain vfses, but that would be up to the end developer and
their choice of vfs.

----------------------------------------------------------------------

Comment By: Alexandre Ferrieux (ferrieux)
Date: 2008-12-04 14:48

Message:
(I know Donal loves when I try to rewrite history, but...)
May I ask why TIP#210 was accepted then ?
Obviously all these matters cannot be handled more elegantly than by the
script level, which provides the proper abstractions; moreover there's no
real performance catch here.
So what's the point of a C implementation ?

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=110894&aid=2388866&group_id=10894