It may not be acceptable, but with minor modifications it can be.
You must keep the current flowing into the battery within the limits specified by the battery manufacturer (not some random dude editing Wikipedia). For safety reasons, you should also ensure that no single point of failure can cause a high current (beyond the manufacturer-specified limit) to flow into the battery.
The only 3.6V primary cells that come to mind are those made by the Israeli company Tadiran, so just for an example, let's check their application information, page 22:
Maximum reverse (charging) currents are listed in the
Tadiran Batteries Product Data Catalogue. To obtain
full performance reverse currents should be kept below
So there's a "maximum current", which varies by battery type (and may, presumably, affect the battery performance), and a general maximum of 10\$\mu\$A that will not affect the battery performance. The "maximum current" is safety-related, not lifetime related. A PN silicon diode such as a 1N4148 will not leak more than 10\$\mu\$A unless you get it really hot (over 100°C). A 200mA Schottky might be okay. But, that's only one constraint.
The Batteries Product Data can be found here. Page 4 contains safety information including the maximum reverse current by battery type (15mA to 100mA), and a number of other cautions.
According to the UL recommendations you should use two series diodes (in case one fails) OR a diode and a current-limiting resistor that will ensure the maximum current is never exceeded, even if the diode fails. That resistor could be just a few hundred ohms (or less, depending on battery type) and meet the requirements of the Tadiran cells listed. For a memory backup or other very low current situation (low tens of uA), the resistor is probably the best because it will drop little voltage compared to a diode.