So strange, Konqueror (this is 3.5.8) worked perfectly until 3.5.6….
So sad, two of my favorite websites gets corrupted…

The alternative may be using one Google Calendar service, but if you want everything integrated it’s disappointing.

There is a project, Mozilla Calendar which has two branches: Sunbird and Lighting, one is a standalone application (useless for me!!), the other is a nice Thunderbird extension.

The bad part is that Mozilla provide Windows, Linux and Mac builds only for i686 systems. So amd64 and newest Intel users (x86_64) cannot use it.

Or better they must use Thunderbird 32 bits binary + Lighting extension build by Mozilla.
So bad.

I’m a Gentoo user, whole system is 64bit and i refuse to have 32bit binaries.
So as Gentoo does not provide an ebuild i compiled it manually:

1) download sources
2) unpack them: tar xjf lightning-sunbird-0.7-source.tar.bz2
3) cd mozilla
4) ./configure –with-system-nspr –with-system-nss –enable-default-toolkit=gtk2 –enable-xft –disable-freetype2 –enable-application=mail –enable-extensions=default,lightning
5) make

you will find Thunderbird binary in the ./dist/bin dir, while our Lighting release is in dist/xpi-stage.

If you want to use my build, i just uploaded the two versions:
Mozilla Lightning 0.7 64 bit version
Mozilla Lightning 0.7 64 bit version with WCAP (Sun Java System Calendar Server) support

Works great on my system, let me know if you have any issue

What else…waiting for the upcoming wordpress 2.3 release and i will update both blogs.

I think that’s all.
And i’m sure you don’t care

see u

Let’s hope it will keep stable for a while.

bye

much better now, lots of space. I think i’ll be here for the whole 2007….

What else..I’m waiting for adsl line to be up, they told me it may take up to 3 weeks…

let’s see.

I have lots of projects/idea in my mind, which i can take care of only during the weekend, but if i don’t have
internet access it’s pretty hard.

The first is to shutdown Blight Zone as it’s unused now.
Next i’ll talk with Zed (XChat maintainer) because 2.8.0 release introduced a systray, so i’ll ask him if he wanna
integrate my plugin in the main branch.

The main idea is to create a forum/blog with my friends where to flame and discuss of everything.
Let’s see how much support i’ll find.

Ciao!

Answer is “no”, of course. Confirmed by kernel guru and mainteneirs, by intent..
should be defined criminal intent…but..ok..
I have no will to open a flame about this, compared to windows stability
(of interfaces i mean, of course!).

In the meanwhile, enjoy the submount patch for 2.6.19, it seems that just a field in the
inode structure has been removed (if i’m wrong… sorry, i had to look into a 32mb patch file, had
no connection, so i don’t know the motivation nor the log of the change).

Tested, works ok on my floppy/cdrom driver.

As usual, enter the subfs-0.9 directory and apply with

patch -p0 < /path/to/the/downloaded/patch

and recompile as usual. I'm going to submit a Gentoo bug for that too.

However this is the listing
--- subfs.c 2006-12-17 16:32:19.000000000 +0100
+++ subfs.c 2006-12-17 16:35:51.000000000 +0100
@@ -310,7 +310,9 @@
if (ret) {
ret->i_mode = mode;
ret->i_uid = ret->i_gid = 0;
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,19)
ret->i_blksize = PAGE_CACHE_SIZE;
+#endif
ret->i_blocks = 0;
ret->i_atime = ret->i_mtime = ret->i_ctime = CURRENT_TIME;
ret->i_fop = &subfs_file_ops;

Download patch

Enjoy and report if you have problems

I tried to use some multimedia applications like MPlayer and all films were not fluid,
audio jittering and so on….cpu usage always at 100%…
During some internet investigation i found some optimization tricks about MPlayer, and after recompiling
it manually from sources with cpu-specific flags i found it sooooo faster,
films were fluid with only 50% cpu usage!

So i thought: why not recompiling the whole system to have more speed?
I started manually but i soon realized it was impossible to manage without a package manager.
However the speed gain was important even if not like in MPlayer’s case.
I can say something between 20 and 30 per cent.
(ok, now someone can start a flame about that. On Pentium2 and after on upgraded Pentium3
the performance gain is really important. Maybe not as on fastest cpus, of course, this is a gcc trouble too,
which is not yet ready to do a perfect optimization, Intel compiler is much better but not opensource)

I started to wrote tons of bash scripts, and in the meanwhile i found Gentoo Linux was my distro!

I installed it in December 2003 and i’m still using it on 2 pcs now (p2 was upgrated to p3 500 and a notebook with celeron 650mhz);
i’m really satisfied of both, even if hardware is so old and slow.

Now i’m going to change my desktop with an Intel Core 2 Duo machine.
Desktop life will be different after that! ehheeh

My main problem is now to migrate my desktop data to the new pc.
I know can you do a low level copy of partition, this is just a 20 mins problem.

The main problem is the arch. I need to migrate from x86 to x86_64 and this is a bit tricky.
I’m now studying some solutions to the problem, but as to Gentoo FAQs, it’s impossibile to
migrate, and they suggest to do a fresh install.

I really don’t believe this. Nothing is impossibile.
For me it’s not enought to migrate my home and my /etc config directory, i want to migrate all
flawlessy.

These are my plans at the moment:

1) boot with a 64bits gentoo minimal livecd (50mb cd)
2) download a portage-snapshot and stage3 pack and install them, having a minimal environment
3) with quickpkg create binary packages of critical dev tools such as gcc, binutils, libtool (any other needed? guess not).
4) low level copy of old desktop harddisk –>> will overwrite 64bits software from stage3
5) reinstall back 64bits packages of point number 3, ignoring some profile stuff, uninstalling old x86 ones
6) compile a 64bit kernel with 64bits native tools
7) reboot with the new kernel and not livecd
8) complete profile migration to x86_64 (maybe may be needed in point 5..)
9) emerge -e system and reboot
10) if all starts properly, emerge -eD world

I think it’s an acceptable procedure.
there are some points which i’m insure, but i suppose lots of unplanned problems may pop out.

I read x86 to x86_64 migration is impossible due to differences in the Gentoo installation process,
but i saw noone…..apart normal x86_64 options in kernel and similar arch specific….

Let’s see in some days, i’ll have new pc at the end of december, but having little time
i need to think about all that before

If works i’ll write a detailed howto, be sure

The problem is its power. MS has too much power.
Can do whatever it wants.

Now with upcoming Windows Vista, they are changing so many things, they
are forcing the security products, the entertainement, the business markets.

And, they are cheating users, with 100 versions.
The most funny is the Home-Basic-For-Poor-And-Dumb-Guys.
It’s a Vista version without any Vista features.
It’s an Xp copy (which was 2k copy) with some minor cosmetic changes.
And costs more, as Acer pointed out.

Ending: everyone wants the Ultimate version, which costs so much.
So people will start to hack and use pirate software, as usual.
If you take a look to the license EULA you can see how many limitation you have.
Can’t copy that, can’t use on different pcs more than XXX times, you have to phone to tell MS “ehi i bought it, please let me use it”,
can’t use on virtual machines, can’t tell your mother you’re using it.

So, even if you buy a licence, you may use it illegally if you’re not
a computer and law expert, even if you didn’t want, but just because you didn’t know.
Silly.
So, just be pirate without buying is the normal thinking.
MS, why are you forcing people like that? that’s sad.

All versions, instead, contains “security” products which are a sort
of tracing/spyware stuff.

WGA (Windows Genuine Advantage) stuff has been hacked in 5 mins.
Now every download has some checks, Windows Media Player, Internet Explorer
DirectX and so on…but they are just boring and useless.

MS will force OGA (Office Genuine Advantage) too.
I still see no sense, maybe they have wonderful stats about that.
Probably the stats are provided by the same companies that says
Windows TCO is lower thatn Linux one.

However, Vista has a very positive thing:
all pc manufacters will have to increase the basic configuration.
Now, if you just want to check your mail, you need at least a dual core pc
with 2 gb of ram, huge harddisk and a powerfull electricity-sucker-gpu.
(DirectX 10 graphic cards seems to use more that 200 Watts…)
Manufacters must update and keep similar prices.

Upgrade to Linux, this is my suggestion.

However, wondeful boxes

Yes, this year I partecipated so little to the community life.
Of course i mean opensource community. What else!

On 28 October, in whole Italy guys have spread Linux knowledge, spotting it everywhere
with the opensource culture and movement.

I was missing. I’m sorry.
Few excuses, just some egoism I guess, i was caring my businness working abroad (

I saw even the TV news were talking about Italian LinuxDay. Don’t give up.

Thanks guys.

Sunday afternoon, after the fantastic Schumacher victory (ehi Alonso, what a long face you had on the podium!), as promised, i took a look to submount and kernel 2.6.18.

What’s up? Since 2.6.18-rc5 (i hope i didn’t confuse the release candidates i checked!)
there has been a change in the function (get_sb_nodev(…) ) to get the superblock, needed by submount.

Now the same function gets an addictional parameter and returns an integer, previously returned the needed superblock pointer.
This is the new prototype (in super.c)

int get_sb_nodev(struct file_system_type *fs_type,
int flags, void *data,
int (*fill_super)(struct super_block *, void *, int),
struct vfsmount *mnt)

mnt is the extra argument i was talking about few lines above.
Compilation fix is not that hard…just pass that extra structure pointer, it will be filled in
simple_set_mnt() function (in namespace.c) and you’ll have the pointer with some valid data.

All seems funny and easy, module compiles and modprobe works, but when i try to mount a device (ex. cdrom) i get a nice (?!?!?) oops

subfs 0.9
BUG: unable to handle kernel NULL pointer dereference at virtual address 00000014
printing eip:
c015a805
*pde = 00000000
Oops: 0000 [#1]
Modules linked in: subfs slamr psmouse
CPU: 0
EIP: 0060:[] Tainted: P VLI
EFLAGS: 00010a12 (2.6.18 #13)
EIP is at do_add_mount+0×61/0xe8
eax: 0000000c ebx: d088ac00 ecx: d1fee220 edx: c12ad050
esi: c03ea2d8 edi: cf857f24 ebp: 00000000 esp: cf857e04
ds: 007b es: 007b ss: 0068
Process mount (pid: 9304, ti=cf856000 task=c12ad050 task.ti=cf856000)
Stack: 00000000 00000000 cc1b8005 00000015 c015b3a9 d088ac00 cf857f24 00000000
00000000 d088ac00 fffffffe 00000000 0000000b d06bc006 c015a6b0 d1fee248
c03ea2d0 cf857ef8 fffffffe c0151577 d1fee220 d06bc000 d1eef534 d1fee220
Call Trace:
[] do_mount+0x5ab/0×601
[] mntput_no_expire+0×14/0×56
[] link_path_walk+0xb0/0xbb
[] filemap_nopage+0x9d/0x2da
[] filemap_nopage+0x16d/0x2da
[] do_path_lookup+0×195/0x1b0
[] __user_walk_fd+0×37/0×41
[] __handle_mm_fault+0x18d/0×675
[] get_page_from_freelist+0x6f/0x2b6
[] __get_free_pages+0×25/0x3e
[] copy_mount_options+0×28/0x10b
[] sys_mount+0×73/0xae
[] sysenter_past_esp+0×56/0×79
Code: 4f 04 25 00 e0 ff ff 8b 10 8b 41 64 3b 82 48 04 00 00 75 61 8b 41 14 3b 43 14 75 0c 8b 41 10 be f0 ff ff ff 3b 07 74 76 8b 43 10 <8b> 40 08 0f b7 40 28 25 00 f0 00 00 3d 00 a0 00 00 74 5b 8b 44
EIP: [] do_add_mount+0×61/0xe8 SS:ESP 0068:cf857e04
<6>subfs exiting.

For me the code seems ok, but i’m not a kernel hacker, so i’d like more feedback and help.
At the moment i’m locked here.

Here is the patch listing and download link:

Download patch (not working properly)

— subfs.c 2006-10-01 07:57:40.000000000 +0200
+++ subfs.c 2006-10-01 07:55:28.000000000 +0200
@@ -412,6 +412,11 @@
struct subfs_mount *newmount;
int ret;

+#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,18)
+ struct vfsmount *mnt;
+ if (!(mnt = kmalloc(sizeof(struct vfsmount), GFP_KERNEL)))
+ return ERR_PTR(-ENOMEM);
+#endif
if (!(newmount = kmalloc(sizeof(struct subfs_mount), GFP_KERNEL)))
return ERR_PTR(-ENOMEM);
newmount->req_fs = NULL;
@@ -429,7 +434,13 @@
strcpy(newmount->helper_prog, SUBMOUNTD_PATH);
if ((ret = proc_opts(newmount, data)))
return ERR_PTR(ret);
+#if LINUX_VERSION_CODE > KERNEL_VERSION(2,6,18)
+ if ( (ret = get_sb_nodev(fst, flags, data, subfs_fill_super,mnt)) !=0)
+ return ERR_PTR(ret);
+ newmount->sb= mnt->mnt_sb;
+#else
newmount->sb = get_sb_nodev(fst, flags, data, subfs_fill_super);
+#endif
newmount->sb->s_fs_info = newmount;
return newmount->sb;
}