I agree about the 2232 (or 2230). Those are my go-to oscilloscopes for general work and they have peak detection which the low end Rigols still lack. When I need real frequency and pulse measurements that may or may not be gated, I break out the 2247A or 7D15. When I need automatic measurements, a pretrigger record, low repetition rates, or waveform storage, then I use a DSO although the 2232 can do all of that except automatic measurements.
For everyday use, I wouldn't trade any of these mentioned above for a new rigol or even one of those Tek junk scopes with 2kB of CCD-memory. Reasons for that are:
-Instant beam, no boot time, no stupid menus, there's a knob for everything so much faster to use
-you see what's really there no ADC artifacts, aliasing or undersampling
-usually displays and triggers on signals well beyond their specs, do that with a digital one
-analog stuff is relatively easy to repair and rarely fails compared to (new) digital circuits and I like to keep my equipment working myself, furthermore the documentation is usually excelent compared to new equipment
-a plugin for about everything in the 7k series like the 7A13 (my everydays favorite since it combines voltmeter and scope); ADC for slow non repeating signals (7A22) in the 7854
The Tektronix CCD oscilloscopes or at least the early ones with peak detection are pretty good and compare favorably with modern "low noise" DSOs once you get past their non-graded displays. I would prefer a short record length with higher waveform acquisition rates than a long record length although the Tektronix CCD designs are slow. Where they suffer is aliasing caused by interleaving of the digitizer but modern DSOs also have that problem to one extent or another if they are implemented that way.
The user interface is generally better; I do not have room on my workbench for another keyboard and mouse. They show a more realistic representation of the signal without a bunch of different modes with inadequate documentation.
I like the 7A13 because it is a great differential input amplifier with slideback capability which puts modern DSOs and most differential probes to shame.
As far as the noise issue goes, look how much trouble Dave had to go to to compare them. That is *not* an advertisement for ease of use or understanding of DSOs. Even worse, too many low end DSOs lack intensity grading which both makes them look worse *and* make them useless for measurements where visual noise matters. The ones that do support intensity grading often do a poor job of it.