emerge -avu portage

Total: 6 packages (2 upgrades, 3 new, 1 in new slot), Size of downloads: 18,646 kB
Conflict: 3 blocks (3 unsatisfied)


Cannot upgrade, some blocks are present. Finding out the minimum version of bash required is simple:


cat /usr/portage/sys-apps/portage/portage-2.1.10.11.ebuild | grep bash

>=app-shells/bash-3.2_p17

To break that circle, lets mask newer bash first.

echo  ">=app-shells/bash-3.2_p39" > /etc/portage/package.mask

New test on emerge

emerge -avu portage

Total: 5 packages (1 upgrade, 3 new, 1 in new slot), Size of downloads: 12,188 kB
Conflict: 2 blocks (2 unsatisfied)

Bash block is gone.

Finding out what is the minimum required python version is a bit harder. Looking at  /usr/portage/sys-apps/portage/portage-2.1.10.11.ebuild we can find a row declaring:

build? ( || ( dev-lang/python:2.7 dev-lang/python:2.6[threads] )

so lets mask python 2.7

echo  ">=dev-lang/python-2.7" >>/etc/portage/package.mask

Please notice the >> this time we append to package.mask. Single > will just overwrite it.

Still no luck:

These are the packages that would be merged, in order:

Calculating dependencies... done!
[ebuild  N    ] dev-libs/libffi-3.0.9-r2  USE="-debug -static-libs -test" 715 kB
[ebuild  N    ] app-misc/mime-types-8  12 kB
[ebuild  N    ] virtual/libffi-0  0 kB
[ebuild  NS   ] dev-lang/python-2.6.6-r2 [2.3.6-r4, 2.4.4-r9, 2.5.4-r2] USE="gdbm ipv6 ncurses readline ssl threads (wide-unicode) xml -berkdb -build -doc -examples -sqlite -tk -wininst" 10,838 kB
[ebuild     U ] sys-apps/portage-2.1.10.11 [2.1.6.13] USE="(ipc%*) less%* -build -doc -epydoc -python2% -python3% (-selinux)" LINGUAS="-pl" 0 kB
*** Portage will stop merging at this point and reload itself,
then resume the merge.
[blocks B     ] >=dev-lang/python-2.6.6:2.6 (">=dev-lang/python-2.6.6:2.6" is blocking sys-apps/portage-2.1.6.13)
[blocks B     ] <sys-apps/portage-2.1.9 ("<sys-apps/portage-2.1.9" is blocking dev-lang/python-2.6.6-r2)

Total: 5 packages (1 upgrade, 3 new, 1 in new slot), Size of downloads: 11,564 kB
Conflict: 2 blocks (2 unsatisfied)

Portage 2.1.9 is still needed by portage 2.6.6. Is there something older?

ls /usr/portage/dev-lang/python/ | grep ebuild

python-2.4.6.ebuild
python-2.5.4-r4.ebuild
python-2.6.6-r2.ebuild
python-2.6.7-r2.ebuild
python-2.7.1-r1.ebuild
python-2.7.2-r2.ebuild
python-2.7.2-r3.ebuild
python-3.1.3-r1.ebuild
python-3.1.4-r2.ebuild
python-3.1.4-r3.ebuild
python-3.2.2.ebuild
python-3.2-r2.ebuild

Nope. the oldest ebuild for python 2.6 is 2.6.6-r2, which requires portage that we do not have. do we have an ebuild for portage that is satisfied with python we do have?

ls /usr/portage/sys-apps/portage/ | grep ebuild

portage-2.1.10.11.ebuild
portage-2.1.10.18.ebuild
portage-2.1.10.19.ebuild
portage-2.1.10.3.ebuild
portage-2.1.6.13.ebuild
portage-2.1.6.7.ebuild
portage-2.1.9.42.ebuild
portage-2.2.0_alpha51.ebuild
portage-2.2.0_alpha58.ebuild
portage-2.2.0_alpha59.ebuild
portage-9999.ebuild

we are running portage 2.1.6.13, there is only one ebuild between 2.1.10.11 and 2.1.6.13. Lets try 2.1.9.42.

emerge -avu =sys-apps/portage-2.1.9.42

Total: 5 packages (1 upgrade, 3 new, 1 in new slot), Size of downloads: 12,310 kB
Conflict: 2 blocks (2 unsatisfied)

Still no luck. The next thing we can try is to ignore dependencies. They are marked in ebuilds for a reason but with some luck we can ignore them and maybe get something semi-working out of it.

emerge -avu --nodeps =sys-apps/portage-2.1.9.42

Another check halts progress again.

This version of portage requires at least python-2.6 to be selected as the default python interpreter

Lets try to force python then.

emerge -avu =dev-lang/python-2.6.6-r2

without forcing python fails like so:

These are the packages that would be merged, in order:

For more information, see the MASKED PACKAGES section in the emerge
man page or refer to the Gentoo Handbook.
(dependency required by "dev-util/pkgconfig-0.25-r2" [ebuild])
(dependency required by "dev-lang/python-2.6.6-r2" [ebuild])
(dependency required by "=dev-lang/python-2.6.6-r2" [argument])

EAPI version depends on portage. Basically, newer portage is needed. Back to the beginning then.

This popt library is needed by pkgconfig. What is the minimum of pkgconfig needed by pyhton?

cat /usr/portage/dev-lang/python/python-2.6.6-r2.ebuild | grep pkgconfig

dev-util/pkgconfig

We might be in luck just a pkgconfig is needed. Any version.

We do have pkgconfig 0.22 /var/db/pkg/ gives a hint here. Mask newer ones.

echo  ">dev-util/pkgconfig-0.22" >> /etc/portage/package.mask

Trying python again:

emerge -avu =dev-lang/python-2.6.6-r2

Total: 15 packages (9 upgrades, 4 new, 2 in new slots, 2 uninstalls), Size of downloads: 37,583 kB
Conflict: 4 blocks (2 unsatisfied)

Blocking again, but now we do have a list of packages that python 2.6.6-r2 wants to install. Lets satisfy those dependencies we can and care about then. Specially those that do more than -r (revision) version upgrades and that do some uninstalls. python-updater is a no go, as it needs newer python of course.

I do use oneshot here as those packages are dependencies of something else anyway and I do not want to add them in my world file.

emerge -avu --oneshot libtool

Libtool was the only one that did work actually and it brought some dependencies along. Good.

Now lets try python with nodeps

emerge -avu --nodeps =dev-lang/python-2.6.6-r2

Unfortunately build fails.

error: /ffi.h: No such file or directory

There actually is no file named so in the whole machine. Quick look around in another, up to date machine reveals that ffi.h is supplied by dev-libs/libffi and actually virtual/libffi was one of the dependencies on python. This tends to happen if nodeps is used on installing packages.

Installing libffi worked and we got a newer python, installed with –nodeps though.  Portage upgrade worked after that also.

the next logical step is to remove package masks we added.

rm /etc/portage/package.mask

Now we can install python with dependencies. This is an important step, because portage can not possibly know, that we installed something with –nodeps and now it might not work as intended.

This takes care of upgrading portage. Next other development tools, python (it got updated already) , perl, gcc should be upgraded. then system, then world

There is no direct route to take if upgrading old Gentoo systems. Just a lot of patience. First upgrade, then mask if upgrade does not work, then –nodeps if mask is not enough. The last thing, if –nodeps also fails is to copy some newer binaries from installation CD. I have had to use that only once fortunately.

The thing to remember – keep Your systems up to date – it is a lot easier this way.

What website statistics package is better? Awstats is more geared towards website administrators and Adsense towards content management. Each has its unique features and usages. I’ll keep my custom Awstats scripts but I tend to look at Adsense more and more.

Lisaks sellele sai tehtud veel mõned töötlused, mis kuhugi ei jõudnud. Üks versioon HDR-ist selle pildi kallal on selline . Siin aga muutusid eriti halvaks varjud auto all – alavalgustusest tulenevad täpid on vägagi häirivad, kuigi samas selle pildi pilved meeldivad mulle kõige rohkem. Kokkuvõttes meeldib mulle aga selle pildi töötlustest kõige rohkem hoopis järgnev pime pilt. Rattakoobaste servad ja ka aknaklaasi hele post on sellised, pilkupüüdvad. Seda pilti ma aga konkursile ei esitanud.  Liiga töödeldud juba.

Tegelikult meeldis mulle sellest fotosessioonist rohkem hoopis pilt mis kalendrisse ei pääsenud ja samast seeriast on ka mõni ilusam HDR töötlus olemas.  Nendest ehk mõni teine jutt.

• Disregard the first reading. It seems that the first sensor reading after a couple of seconds is highly fluctuating.
• Use lower voltage. You loose some some sensory range, but get more precise results. Specially if You know that the range is not needed anyhow. Normal room temperature will rarely get to 100°C for example.
• Get more than one reading. It gives a better virtual resolution and fluctuations get smoothed out.
• Have a small delay between readings. This keeps fluctuations down.
• Test Your system before putting it into production. Theory might need to be adjusted to work in real world.

For my heater controller project I made this small function to read my LM35 sensors:

// read analog LM35 sensor
float LM35_temp(int pin){
unsigned int tempread = 0;
unsigned int intmp;
float temp;
int i;

Serial.print("Reading analog temp on pin: ");
Serial.println(pin);

// first sample seems to fluctuate a lot. Disregard it
intmp=analogRead(pin);
Serial.println(intmp);
delay(60);

// 11 bit virtual resolution, according to http://www.atmel.com/dyn/resources/prod_documents/doc8003.pdf
// arduino ADC is 10 bit real resolution
// for 12 bit resolution 16 samples and >> 4 is needed
for (i=0;i<=3;i++){
intmp=analogRead(pin);
tempread = tempread + intmp;  // read  more samples. better stability.
Serial.println(intmp);
delay(60);
}
tempread = tempread>>2;

// default voltage is 5.0, analogReference(INTERNAL) sets it to 1.1
// temp = temp / 5.0;
//temp = (5.0 * (float)tempread* 100.0)/1024.0; //convert the analog data to temperature
temp = 110 * (float)tempread / 1024; // analogReference(INTERNAL); needed
Serial.print("Temp is: ");
Serial.println(temp);
return temp;
}// LM35

Also analogReference should be set, preferably in setup

// analog sensors reference.
// more precise but smaller range
analogReference(INTERNAL);

Writing code on web page is not the most effective way, so the same code can also be downloaded from here.

Steps to make next and previous post links category sensitive:

Locate post template. File named single.php usually, if all post use one, default, template.

Locate  previous_post_link(‘%link’, ‘%title’) in that post template and modify it to look like  previous_post_link(‘%link’, ‘%title’,TRUE).

Repeat it for next_post_link. That is it.

First and second argument to _post_link must not be %link and %title, but they usually are. Leave them as they are.

Link to wordpress documentation is here.

Most of my traffic comes from search engines, mainly from google, which usually display some excerpt of web page text, so there should be something that made people click on those pages. What made them leave when they see the full page? Either the excerpt was misleading or they got all they needed from that single page and my webpage had nothing more that might interest those visitors. Looking at visitors from google search, my bounce rate is about 80% which is clearly wrong. How can I get it down?

There are already a couple of things I did.

• Firstly I looked at main landing pages and added links to other articles at the end of those posts.
•  I also installed a related posts plugin.   On average, using insufficient data really, bounce rate got down by about 10% for using related posts plugin. Links to other articles at the end of posts brought bounce rate down by additional 5%.
• Another thing about links – visitor should be able to spot them. My default theme had grey text on white background and links were a bit different shade of grey.  I did select another color scheme. Links are  pale blue now, which should be easier to spot.
• I also modified next and previous links on articles to be category dependant as described here!
• Social interaction is also an important thing. Adding Google +1, Facebook, Twitter etc. links is a start. 42functions Social Media Tracking plugin is one way to to it.

What should I do next? More interlinked content for sure.  The content visitors are looking for can be guessed by looking at site content-> landing pages -> entrance paths. Maybe when visitors are using a post as their landing page and then clicking at the same post on sidebar menu is a hint?

After ten days,  the result is promising. Average bounce rate is down from 72% to 61%, on better days it is about 55%.