Regulator topology design philosophy

Regarding semiconductor based voltage regulator circuits (being fed from typical toroidal dual rectified transformer secondaries), does anyone have any strong opinions as to which is better: to use two identical positive regulator circuits and link them together near the output to get a +/gnd/- supply, or use positive and negative regulator semis.

With my chosen design means the options are to 1) link two identical LM317 with a TIP35C current shunt transistor or go for a more schematically (wow, it really is a word!) symmetrical but less physically self-similar LM317 + 337 + TIP35C PNP + TIP36C NPN.

I've all components, so that's not an issue.

In any event I'm reasonably happy to have worked 'schematically symmetrical' in to a serious conversation...
 
and for the alternative view, I think two excellent identical voltage regulators series connected, could perform better than a series connected complementary pair if one of the complementary outperforms the other due to the use of PNP?Pchannel transistors.

The slight advantage of the complementary is that both dual secondary and centre tapped secondary are suitable.

Two same sex regulators require a dual secondary transformer.