SPECS: wesnoth.spec - fixed install problem with tent.png

Szymon Siwek sls at poczta.wp.pl
Thu Aug 28 14:09:10 CEST 2008


On Thu, Aug 28, 2008 at 11:51:07AM +0300, Elan Ruusamäe wrote:
> On Thursday 28 August 2008 01:20, Szymon Siwek wrote:
> > On Wed, Aug 27, 2008 at 08:33:10PM +0200, Zsolt Udvari wrote:
> > > > What's going on? tent.png looks like proper PNG file. wesnoth builds
> > > > without errors. Is it some problem with Zsolt's filesystem (Macintosh
> > > > HFS, I guess)?
> > >
> > > I don't know what's the problem. E.g. feh and geeqie can show the
> > > tent.png but 'file tent.png' returns the prev error message. I was
> > > wondering too. I'm using reiserfs, so this is very strange :(
> > >
> > > I've unpacked again and same error:
> > >
> > > $ file data/core/images/terrain/tent.png
> > > data/core/images/terrain/tent.png: ERROR: Macintosh HFS Extended
> > > version 61389 data (unclean) vasprintf failed (Invalid or incomplete
> > > multibyte or wide character)
> >
> > OK, I see - file-4.25 is looney and doesn't recognize PNGs.
> 
> no.
> 
> it recognizes file as HFS first, then as PNG, which is wrong imho as PNG has 
> pretty clear mime-magic.
> 
Are you sure? "file *.png" returns "data" for all PNGs on my builder.
BTW I've wrote patch which disables recognizing Macintosh filesystems (it's
insane to use 2-byte magic string)

-- 
Szymon Siwek
-------------- next part --------------
--- file-4.25/magic/Magdir/macintosh.orig	2008-08-27 23:19:21.547928744 +0200
+++ file-4.25/magic/Magdir/macintosh	2008-08-27 23:20:44.042259097 +0200
@@ -272,14 +272,15 @@
 # Alternatively, the boot block is supposed to be zeroed if it's
 # unused, so a simply >0 should suffice.
 
-0x400	beshort			0xD2D7		Macintosh MFS data
->0	beshort			0x4C4B		(bootable)
->0x40a	beshort			&0x8000		(locked)
->0x402	beldate-0x7C25B080	x		created: %s,
->0x406	beldate-0x7C25B080	>0		last backup: %s,
->0x414	belong			x		block size: %d,
->0x412	beshort			x		number of blocks: %d,
->0x424	pstring			x		volume name: %s
+# "\xd2\xd7" is has many false positives
+#0x400	beshort			0xD2D7		Macintosh MFS data
+#>0	beshort			0x4C4B		(bootable)
+#>0x40a	beshort			&0x8000		(locked)
+#>0x402	beldate-0x7C25B080	x		created: %s,
+#>0x406	beldate-0x7C25B080	>0		last backup: %s,
+#>0x414	belong			x		block size: %d,
+#>0x412	beshort			x		number of blocks: %d,
+#>0x424	pstring			x		volume name: %s
 
 # "BD" is has many false positives
 #0x400	beshort			0x4244		Macintosh HFS data
@@ -296,24 +297,25 @@
 #>0x412	beshort			x		number of blocks: %d,
 #>0x424	pstring			x		volume name: %s
 
-0x400	beshort			0x482B		Macintosh HFS Extended
->&0	beshort			x		version %d data
->0	beshort			0x4C4B		(bootable)
->0x404	belong			^0x00000100	(mounted)
->&2	belong			&0x00000200	(spared blocks)
->&2	belong			&0x00000800	(unclean)
->&2	belong			&0x00008000	(locked)
->&6	string			x		last mounted by: '%.4s',
+# "H+" is has many false positives
+#0x400	beshort			0x482B		Macintosh HFS Extended
+#>&0	beshort			x		version %d data
+#>0	beshort			0x4C4B		(bootable)
+#>0x404	belong			^0x00000100	(mounted)
+#>&2	belong			&0x00000200	(spared blocks)
+#>&2	belong			&0x00000800	(unclean)
+#>&2	belong			&0x00008000	(locked)
+#>&6	string			x		last mounted by: '%.4s',
 # really, that should be treated as a belong and we print a string
 # based on the value. TN1150 only mentions '8.10' for "MacOS 8.1"
->&14	beldate-0x7C25B080	x		created: %s,
+#>&14	beldate-0x7C25B080	x		created: %s,
 # only the creation date is local time, all other timestamps in HFS+ are UTC.
->&18	bedate-0x7C25B080	x		last modified: %s,
->&22	bedate-0x7C25B080	>0		last backup: %s,
->&26	bedate-0x7C25B080	>0		last checked: %s,
->&38	belong			x		block size: %d,
->&42	belong			x		number of blocks: %d,
->&46	belong			x		free blocks: %d
+#>&18	bedate-0x7C25B080	x		last modified: %s,
+#>&22	bedate-0x7C25B080	>0		last backup: %s,
+#>&26	bedate-0x7C25B080	>0		last checked: %s,
+#>&38	belong			x		block size: %d,
+#>&42	belong			x		number of blocks: %d,
+#>&46	belong			x		free blocks: %d
 
 # I don't think this is really necessary since it doesn't do much and 
 # anything with a valid driver descriptor will also have a valid


More information about the pld-devel-en mailing list