Is the stuff said about the LED lighting in this article, is true of false?
Link (Original untranslated article) .

Led bulbs with lower color temperature would help, but you can't get away from the blue LED chip making the phosphor make light.
I think you can only get so low with LED because the lowest wavelength (deep red, from phosphor, if possible) and the blue LED chip and that you cannot make more light than was put into the phosphor by the chip.
A better solution would to be to use RGB LED lights to make white light.

Led bulbs with lower color temperature would help, but you can't get away from the blue LED chip making the phosphor make light.
I think you can only get so low with LED because the lowest wavelength (deep red, from phosphor, if possible) and the blue LED chip and that you cannot make more light than was put into the phosphor by the chip.
A better solution would to be to use RGB LED lights to make white light.

If I ever get the money, I would like to try and build an LED streetlight that uses RGB leds over white ones. They would have to be diffused somehow as to not make colored shadows which might make some sick.

I don't think RGB LED would help. I've seen the white light that produced by color changing LED lights, and it is horrible as well.
I think that the only way to prevent lighting pollution, is to return to mercury and LPS lighting. You can also use HPS lanterns with cutoff optics, but I suspects that the reflected light from the earth also plays role in making lighting pollution so cutoff optics isn't a 100% solution.
To prevent lighting pollution, LED lighting shouldn't be used in the first place, because its spectrum contains the greatest amount of blue light compared to other white light sources and sodium based lamps.

What about high CRI Violet LEDs?
How light polluting would that be?

To have a white LED with a high quality white light, the die should be UV to allow more phosphors to be used. This method is less efficient than the blue die.

Light pollution is in the first place the matter of a not so good optics and the popularity to illuminate everything around, every building or so. Then all cities look nearly like Vegas Strip combined with a Disneyland- something I really do not like at all (except of really the genuine Las Vegas strip and Disneyland parks).
I don't know who made that observation (it came from the end of 19'th century and was about the coal usage vs the efficiency of the technology), but it really applies for the lighting as well: The more efficient technologies haven't made the needed lighting to consume less power, but it caused an explosion of illuminating really everything. And not because the visibility is needed there, but just because someone think it "looks better". That is really the main wasting of the natural resources (the darkness at night belong to them), where the environment protection should start (when speaking about lighting), not banning the incandescents at homes. Why the heck all the illumination, when there is anyway no one outdoor to enjoy it (most villages with "nicely illuminated churches").

The color is of second importance. If a SOX shines everywhere (because that lamp size just does not allow any better), it does not matter the astronomers are easily capable to filter that out, when the sky is just orange haze (example from Flagstaff, AZ; if there were no such wasteful installations and really only the necessary street lighting, the observatory won't be disturbed even when they would use broadband MH's).

What about high CRI Violet LEDs?
How light polluting would that be?

That would be less efficient and I would guess the violet will have similar effect on the sleep.

And regarding RGB:
First the fact it can form very broad color range, the CRI of that is quite poor, not that much above 50 or so (CRI is not about what color the light emits, but how well it renders the colors of illuminated objects; and there the missing gaps between the Red, Green, and Blue components are responsible for the poor color rendering).
Second, the green LED's are the least efficient among all LEDs (even when the human eye is the most sensitive in that wavelength region, the greens still have the lowest efficacy). It is the wavelength range, where both of the two major chemistry families have very poor results. So poor, more efficient is to use a phosphor illuminated by a blur chip, if the leaking blue would not be a problem (white LED's for illumination; for indicators and color panel application the color cleanness is more important, so the lower efficacy is accepted).

They are in the process of changing over all the cobra heads in my area to LEDs.  I expressed my concern about the LEDs being too bluish and the power company agreed with my concerns and when approaching the various fixture manufacturers as well as lighting engineers they made that perfectly clear.  They found a manufacturer who presented them with a good solution.  I've been driving out in the community here every night as they get about 30 heads done a day and it is really neat seeing the difference with LEDs on one side of the road and HPS on the other.  The color they tell me is between 4000-4100k.  It truly is more of a cool white color vs the crisp white.  It's perfect. 

Similar light-pollution stories in the news today. The milky way can be glimpsed but cannot be seen in all it's glory by nearly all humans.

According to my Sky Quality Meter, the skyglow over my city has been slightly lower in the last few years, mostly because of lumen depreciation as far as I can tell. If LED conversions take place quickly, light-pollution will at first rise since the same (or sometimes higher) levels on the ground are required or desired. When LED lumen depreciation takes hold, it might get darker.

Melanopsin cells in our eyes which chronically suppress melatonin levels, have a peak spectral sensitivity near 460nm which is also near the peak of the blue generating InGaN LEDs. The spectral range of these cells seems to span the wavelengths of 459-485 nm (just the blue!). Apparently, violet light which spans say about 390-430 nm is excluded or was absentmindedly omitted in the studies.

Similar light-pollution stories in the news today. The milky way can be glimpsed but cannot be seen in all it's glory by nearly all humans.

According to my Sky Quality Meter, the skyglow over my city has been slightly lower in the last few years, mostly because of lumen depreciation as far as I can tell. If LED conversions take place quickly, light-pollution will at first rise since the same (or sometimes higher) levels on the ground are required or desired. When LED lumen depreciation takes hold, it might get darker.

Melanopsin cells in our eyes which chronically suppress melatonin levels, have a peak spectral sensitivity near 460nm which is also near the peak of the blue generating InGaN LEDs. The spectral range of these cells seems to span the wavelengths of 459-485 nm (just the blue!). Apparently, violet light which spans say about 390-430 nm is excluded or was absentmindedly omitted in the studies.

So violet should be good for Melatonin?
And I thought RGB LED lights should have a good CRI because Triphosphor, which has the same concept, has a good CRI.

So violet should be good for Melatonin?

No. The influence of violet was just not subject of any study. That means no one knows.
The studies took into account only the cited range, I guess just because they want to somewhat quantify the acceptable "safe levels" of the blue peak from the LED's. And they did care only about the LED's, because till the study have some real impact on related regulatory standards, they expect from the known viable technologies the LED's will be the dominant one in wide use, mainly on new installations (only those may be impacted by the eventual new regulations).

And I thought RGB LED lights should have a good CRI because Triphosphor, which has the same concept, has a good CRI.

The "tri-phosphors" indeed emit just three spikes, but the products using them have at least four or even 5 spikes. You should not forget the "natural" mercury radiation, which is quite important as well (some mixes have one of the phosphor peak the same as one of the mercury lines, therefore some products have only 4 bands; thats the reason for the differences among makers).
The extra lines, even when they are not that strong (and carry only e.g. 10% of the lumen output) are very important for the final color rendering (with just three lines, there are large gaps between, so if some object color component reflects only narrow part of the spectrum, it is more likely to "fit between" them and so being not visible; with more spikes the gaps are getting narrower, so you would need way more narrower band object color to get poor rewndering and a chance of that gets really steeply reduced as the gap get narrower and narrower)...

There is a strong difference when implementing a light source for an illumination vs capable of wide light color range.
The first needs to eliminate the wide gaps in the spectrum so wide band primary sources are ideal.
The second needs to cover as large part of the "horseshoe" color chart, with RGB it means a triangle between the color positions corresponding to each of the individually controllable component (you may notice the RGB may cover only about half of the diagram area; hovewer the natural colors, which are supposed to be displayed when it matters, are mostly around the center, so within the triangle). If you want to boost the color ares, you need more light components (saturated yellow is not possible to display with just RGB, as the saturated yellow does not stimulate the blue sensing cells, but the green component from the RGB does so)

So violet should be good for Melatonin?

Looks like it's a big unknown. (Medved posted a similar response)

The studies made with blind people (to see if they can change their circadian clocks) made references to testing with red, green & blue light. The blue light had the effect in changing sleep patterns of blind people, discovering that there are extra cells, the melanopsin cells that regulate sleep patterns which are different from rods & cones used for vission. It seemed like the researchers were not concerned with spectral precision, and some one else may have picked up that "blue" light is the color that supresses melatonin production so we got that 459-485 nm range. I searched in vain to find if violet light was tested and I came up with nothing. They may have done it by now, or are actively trying to find as to which spectral regions melanopsin cells are precicely affected. Or they should be doing this.

Need to ask members of LG...

Since virtually all LED heads are FCO, this means there's zero direct up-light.
Since many thousands of HPS heads with sag lens have (what is it) about 3 to 5% up-light?

So if all HPS heads are replaced with LED FCO fixtures, then there sould be a 3-5% drop in light pollution. Ah, but what about reflections from the ground you say?

I measured a variety of lumen levels on ground from street lights and parking lot lights. My conclusion is forecoming.

LED will increase the light pollution regardless of the optics, as it is very rich in blue compared to other lightsources.