in my opinion people need : how long working my raspberry pi or indicator in percentage (numbers generaly) not voltage.

users need add many different kind of power container, 

I would like to buy one box and put my AA accumulators, lipo or car accumulator to one box

and using router and small computers.

Ask neightbor about usefull Your idea

I am still thinking about doing something similar. I've bought one of those chinese usb powerbanks, with 10000 mAh. Of course it stopped working almost immediately. My plan is to have a microprocessor that does the following, once per second:

1 Loop:

1.1 set charge current, put in measurements table column 1

1.2 measure input voltage (to charger), put in measurements table column 2

1.3 increase charge current

2) multiply I*V (column 1 * column 2) to calculate highest charging power.

3) set optimal charging current for the next second.

This is because PV panels have an optimal load at which they provide power, and this optimal load is dependent on how much sun they get at the moment, which could change from second to second.

The charge current could easily be set, for instance using this charger:

http://www.farnell.com/datasheets/77596.pdf

by just varying the programmable current. This can be as crude as using a DAC to drive a NPN transistor, which connects to the PROG pin, or even DAC made from some output pins with a resistive network.

Thoughts?

It's better, but wastes all the energy above the battery voltage.

A better solution is to step up or step down the voltage of the cell and charge the battery. You got a wider operation window for the MPPT and less power loss.

I put the charging, switching and regulation stuff all in a small attiny (8 Pin) device. I do not need any expensive switching and charging IC:
https://sebastianfoerster86.wordpress.com/2016/01/02/attiny13-solar-cell-charger-hardware/

I don't think that a buck/boost converter will result in better energy performance. As long as PV panels are used that are 5V, otherwise it doesn't work very well.

Assuming that;

The PV cell will sag under load. In my solution, by changing the charge current the PV panel will result in varying output voltage, and so varying output energy delivery, until you arrive at the load point with the maximum power generated; the optimal power point.

In your solution with the buck/boost converter you would have to do exactly the same thing; vary the required output current (which your didn't mention, so I assume you didn't). Otherwise you won't do power tracking of the PV cell, and it will sag to a point that is not optimal.

Ficticious example: say that the solar panel has an optimal point of 200 mA by 4.3 V at a certain light intensity. Any lower current draw and the PV voltage will go up but energy delivery will be less. Any higher current draw and the PV voltage will go down and energy delivery will be less. If you attach a buck/boost converter, and you set it to plain 4.2 V, and you just let it do it's work, and having a fixed maximum output current of 300 mA, then it's guaranteed that the power delivery will be suboptimal. Again, that is because the power setting will be somewhere else than 200 mA at 4.3 V.

Whereas, in my solution, if you just program the battery current to be variable, you'll get the the most optimal power setting for the panel. 

For light intensities below the battery voltage, it is true that a boost converter may generate higher voltages. But hardly any energy, and unless you can control the charge current, no energy at all because the voltage will drop below the threshold of the boost converter.

Secondly, less power is wasted because a buck/boost converter also uses energy.

Finally, charging IC's are not expensive. Here's one for £1.82 from ebay with programmable current:

http://www.ebay.co.uk/itm/IC-LI-ION-LI-POLY-CHARGER-SOT23-5-Part-MICROCHIP-MCP73812T-420I-OT-/351562267201?hash=item51dabe7a41:g:anoAAOxy7nNTPxdo