Wednesday, September 11, 2013

Redshift to get some relief for your eyes

I am working in front of computer all day long, and looking at the bright screen hurt my eyes.  I was relied on the plugin ToggleDocumentColors to give me a grey background and black text with browser.  But it does not look good on some web sites and it only works on browser.  Then I came across this Redshift program, and it has a script.  After installing it with sbopkg, found my latitude and longitude and fire up the program, instantly I could feel the stress of my eyes are reduced.

It may take sometime to get used to it as the color on the screen has a little reddish tone.  Unless you are working on graphics or color sensitive work, normal interaction with the web site or application has no problem at all and your eyes got some relief.  Since the discovery of this little tool, I have installed it to all my systems.

It has an GUI front end gtk-redshift that comes with the package.  It will display an icon on your system tray area, allow you to toggle it by a single click.

Highly recommend this little tool to anyone that need to work in front of computers.

Friday, August 16, 2013

Automatic new line at program exit for Ada.Text_IO output.

For these two simple programs:

with Ada.Text_IO; use Ada.Text_IO;

procedure Hello is
    Put(" ");
end Hello;
with Ada.Text_IO; use Ada.Text_IO;
with Ada.Text_IO.Text_Streams; use Ada.Text_IO.Text_Streams;

procedure hello_stream is
    std_out   : Stream_Access;
    std_out := Stream (Standard_Output);

    String'Write (std_out, "Hello");
    String'Write (std_out, " ");
    String'Write (std_out, "World!");
end hello_stream;

When executing the two program, you would have two different result, for hello program:

# ./hello
Hello World!

For hello_stream program:

# ./hello_stream
Hello World!# 

I have known this problem for some time, but did not dig into the root cause of it.  Recently I asked this question in comp.lang.ada group for a root cause of it. And Jeffrey gave me the answer.

The related references in ARM are A.7(6) and A.10.1(86).

For gnat compiler File_Type is defined at

type Text_AFCB;
type File_Type is access all Text_AFCB;

type Text_AFCB is new FCB.AFCB with record
end record
procedure AFCB_Close (File : not null access Text_AFCB);
Where FCB.AFCB is a tagged type defined in

In AFCB_Close() procedure, it calls Terminate_Line(), which will try to insert a new line if not terminate properly.

It seems, when program exit, gnat doing the finalization for Standard_Output call AFCB_Close(), which insert the new line if necessary.

p.s. This is tested under Slackware64 Linux with GCC gnat compiler.

Wednesday, July 10, 2013

gvim and i3 not playing too well together

It seems gvim is not playing nicely with tiling manager.  I encountered this bug.

It seems I am not alone with this problem.  To maximize and restore the window each time doesn't seem to be a good workaround.  The workaround I used are from the stackexchange, add the following line into my .vimrc:

set showtabline=2
It always display the tab bar.  Although it wastes some spaces, it gives me a proper displayed window.

Thursday, July 4, 2013

Old PC with a new life in Slackware

One of my home PC died (HP Pavilion Media Center m8530f), and not seeing a great need for a new one, I found my wife's old PC laying in the garage.  It is a brand-less PC with an Asus motherboard P5VDC-MX (an old revision with only 2 RAM slots) couple with Pentium 4 (64-bit enbaled), 1G of DDR RAM, CD-writer and floppy drive!  Oh, it has SATA support also.  That is good as I could reuse the SATA components from the HP PC.  It has been collecting dust for years in the garage, and I was afraid whether it could still boot. 

After cleanup the dust and power up, it run without much problem, its BIOS battery run out, replaced it with the one from the HP PC.  I installed the DVD, multi-card reader and hard drive from the HP PC into this old board. It seems I could not use both IDE channel when SATA is in used (long time to boot and hard drive is not recognize correctly).  I have to disconnect the second IDE channel and have both old IDE drives on the first channel.  Then it boots up fine, even with my previous custom made kernel for the the HP PC.  The on board VGA is too weak, and I solicited an old ATI radeon 9200 SE AGP card from my friend.

After installed everything, I recompiled the kernel to match what I had on the PC, and almost everything work except for the sound.  I always got something like:

hda_intel: azx_get_response timeout, switching to polling mode: last cmd=0x12345678
testing with aplay gave the speaker repeating tone.  After searching the net and consult the documents in kernel source, it seems switching to polling mode is not a serious problem at first glance.  But I don't think it is right. Setting probe_mask=1 for the snd_hda_intel module gave no help.  Then I recompiled the module with debug turn on and I saw lines indicate it has MSI turned on, after setting "enable_msi=0" to off, sound is working.

The display has some minor problem, in console, the text is quite dim and it will become dimmer if running after a mc instance.  One thing is strange that if I highlight a few lines in the console with mouse, the brightness of the text will increase depends on how much line I have highlighted.  In X, I have the same problem if I have an xterm on full screen, it will become dimmer, if I have a relatively bright window opened next to it, it will be fine.  Well it is an old PC and usable, so I don't complaint too much ^_^

P.S. when compiling kernel, I feel the age of this PC ^_^ it took almost an hour.  Running Slackware64 14.0 on this old one.

P.P.S. it seems my problem of dim text in console has gone away, it maybe related to the old radeon card/mainboard hasn't been use for years, after a few days of usage, it seems back to normal.

Friday, June 28, 2013

3 days using i3 tiling window manager under Slackware

I have heard about tiling window manager for some time, but lack the initiative to test one.  Recently I finally got fed up with resizing two windows side by side and decided to give titling manager a test.

After some research from the web, I settled on i3 tiling window manager for its simplicity.  With the help from SlackBuilds, I got the window manager installed and run without hassle (I did recompile the cairo with xcb support (from current) and use the latest i3 release version 4.5.1).

It took me a while to get familiar with the key bindings.  Thanks for the good documentation, I got used to it quickly and customized the configuration to suite my own needs.  No longer do I need to fight with the mouse to place the window to gain the last pixel.  Wow, I wish I should try it earlier.  It just simply awesome! Oh, and awesome is another good tiling window manager too ^_^.

I have used various window mangers and DE during the years of using Slackware.  From the beginning, it was fvwm, fvwm2, IceWM, window maker, GNOME, KDE, XFCE, LXDE, blackbox, EDE, E16, E17, fluxbox.  Now I am simply sold to the tiling manager and won't go back to the normal one.

Thursday, June 13, 2013

Problem with wireshark 1.10.0 in Slackware 14

I just tried to compile wireshark 1.10.0 under Slackware64 14.0, using the build script from SBo.

It compiled fine, but running with segmentation fault.  It seems wireshark itself try to use libnl3 but libpcap in Slackwre is using libnl (version 1).  After adding configuration "--with-libnl=1" to force wireshark to use version 1 of libnl, it does not crash anymore.

But it did not use gnutls, it seems the version (3.0.23) in Slackware is too old, it required at least 3.1.10.  Currently, the latest stable version is 3.2.1, but it requires update nettle to 2.7 also.  I ended up update the gnutls lib to 3.1.12 to compile wireshark with tls support.

It seems Slackware-current does not update gnutls library yet, hope next version of Slackware could update the gnutls version to a newer one.

PS. seems nettle 2.7.1/gnutls 3.2.1/wireshark 1.10.0 works fine after I recompiled the 3 packages.

Tuesday, June 11, 2013

git error message '... does not point to a valid object'

I tried to sync my local git repository to the remote server and got some error messages saying certain objects are not valid object: "... does not point to a vlid object".

After searching the net, these two commands solve my problem:

git remote prune origin
git gc --prune=now

After that, re-sync my repository without problem.

Thursday, May 9, 2013

Google Glass?

Google Glass seems grasp quite some attention.  From the limited information gathered, I don't see it an attractive gadget at all.

I am not speaking from technical points of view, like processor power, battery life, user interface, etc. but its usage point of view.

Maybe for some specific venues or activities, it could a good use, but not in general usage.  The display/notification from it disturb one's attention, it may danger your life, especially when you are driving or walking on the street, your attention got diverted and tragic accident would happen.  I could imagine there would be laws to ban usage of Google glass when driving if it could ever reach that kind of massive market.  Maybe Google should add a small radar into it to prevent collision ^_^

I am not sure how much would it affecting one's vision health when switching your eye focus back and forth from the display to background scene.  Would it have any bad effect on how the brain process vision?