Grokbase
Topics Posts Groups | in
x
[ help ]

Tim Jenness (t.je...@jach.hawaii.edu)

Profile | Posts (188)

User Information

Display Name:Tim Jenness
Partial Email Address:t.je...@jach.hawaii.edu
Posts:
188 total
188 in Perl 5 Porters

5 Most Recent

All Posts
1) Tim Jenness Re: 5.8.9 RC1
| +1 vote
Sorry about that. Dave prodded me about it for 5.10.1 to sync with bleadperl but I ended up...
Perl 5 Porters
[ Profile | Reply to group ] [ Flat  Thread  Threaded ]
On Mon, 17 Nov 2008, Nicholas Clark wrote:

> File;:Temp 0.21 appeared after I'd decided that it was freeze time for 5.8.9

Sorry about that. Dave prodded me about it for 5.10.1 to sync with
bleadperl but I ended up incorporating fixes from some other outstanding
RT bugs at the same time.

--
Tim Jenness
JAC software
http://www.jach.hawaii.edu/~timj
2) Tim Jenness Re: File::Temp / File::Path interaction
| +1 vote
That seems fine with me. I'll put together a new release.
Perl 5 Porters
[ Profile | Reply to group ] [ Flat  Thread  Threaded ]
On Nov 13, 2008, at 9:24 AM, Gisle Aas wrote:

>
> On Nov 13, 2008, at 20:04 , David Golden wrote:
>
>> On Thu, Nov 13, 2008 at 1:03 PM, Gisle Aas <gisle@activestate.com>
>> wrote:
>>> On Nov 13, 2008, at 18:10 , David Golden wrote:
>>>> On Wed, Nov 12, 2008 at 8:43 PM, Tim Jenness <t.jenness@jach.hawaii.edu
>>>> >
>>>> wrote:
>>>>>
>>>>> Should I simply put an eval around the rmtree? Or does that
>>>>> sound too
>>>>> simplistic?
>>>>
>>>> I would go ahead and do that, even if File::Path is fixed since
>>>> people
>>>> might upgrade File::Temp and not File::Path.
>>>
>>> I don't think I would recommend that. File::Path actually tries
>>> to croak if
>>> you try do delete the current directory before it starts deleting,
>>> so if you
>>> just put an eval around it it will just silently fail (under the
>>> right
>>> conditions).
>>
>> That's my point. There are possible reasons why temporary
>> files/directories can't be deleted on various platforms -- I'm most
>> familiar with Win32, where files open files inside the temporary
>> directory prevent removal, etc. This can happen due to virus scanners
>> operating, directory indexing, etc. I even think that I've had
>> directory removal be hosed because the OS hadn't finished deleting
>> files yet. (Hackery: delete files, sleep 2 seconds, delete
>> directory.)  It's ugly stuff.
>>
>> Because File::Temp cleanup is 'magic' -- happening behind the scenes
>> of a program -- for it to be "user-friendly" in my view, it should
>> make a best-efforts attempt to clean up files/folders (assuming it's
>> set to do so), but should not have failure to remove them be fatal to
>> the ongoing execution of the program -- they are, after all,
>> temporary, and eventually leftover temporary files/directories can be
>> removed by the OS or the user.
>
> That's fair enough. I just think it might be a bad idea to totally
> silence the rmtree() errors that might occur. Something like this
> might be better:
>
>   eval { rmtree($dir) }; warn $@ if $@ && $^W;
>

That seems fine with me. I'll put together a new release.

--
Tim Jenness
Joint Astronomy Centre
3) Tim Jenness File::Temp / File::Path interaction
| +1 vote
Debian bug http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=479317 reports that File::Temp and...
Perl 5 Porters
[ Profile | Reply to group ] [ Flat  Thread  Threaded ]
Debian bug http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=479317
reports that File::Temp and File::Path can behave badly if the user  
has chdir'ed into the temporary directory and it's then cleaned up  
either during object destruction or END block. The reason is that  
rmtree aborts if it can not change back to the directory that it  
started from but does not realise that it started from inside the tree  
that it just deleted.

Should I simply put an eval around the rmtree? Or does that sound too  
simplistic?

I noticed this whilst preparing a new release of File::Temp to sync  
with bleadperl.

--
Tim Jenness
Joint Astronomy Centre
4) Tim Jenness Re: ext/Time/Piece/Piece.xs use of strptime
| +1 vote
You have a good point there... :-)
Perl 5 Porters
[ Profile | Reply to group ] [ Flat  Thread  Threaded ]
On Tue, 12 Feb 2008, Jerry D. Hedden wrote:

>> because removing modules always leads to backwards compatibility issues?
>> There are people out there who are not allowed to install CPAN modules and
>> just have a shipped perl. Their software will break spectacularly if
>> modules are removed without a deprecation cycle.
>
> That argument, of course, begs the question of way it was
> removed in 2001.
>

You have a good point there... :-)

--
Tim Jenness
JAC software
http://www.jach.hawaii.edu/~timj
5) Tim Jenness Re: ext/Time/Piece/Piece.xs use of strptime
| +1 vote
because removing modules always leads to backwards compatibility issues? There are people out there...
Perl 5 Porters
[ Profile | Reply to group ] [ Flat  Thread  Threaded ]
On Tue, 12 Feb 2008, Jerry D. Hedden wrote:

>> Or why not contact Kevin Ruddy, the apparent author who
>> contributed it to FreeBSD.
>
> Why is Time::Piece even in core? It was removed way back in
> 2001. Then in 2006, Steve Peters added it back in for no
> apparent reason. It's completely extraneous; nothing in
> core uses it.  Why not just remove it?
>

because removing modules always leads to backwards compatibility issues?
There are people out there who are not allowed to install CPAN modules and
just have a shipped perl. Their software will break spectacularly if
modules are removed without a deprecation cycle.

--
Tim Jenness
JAC software
http://www.jach.hawaii.edu/~timj

spacer
Profile | Posts (188)
Home > People > Tim Jenness