"Standard" glossing (following the Leipzig rules) uses a linear model of breaking down words into morphemes. In other words, it assumes you can draw lines between all the morphemes to separate them.
But what do I do when these assumptions break down?
For example, the Arabic word كتب (kataba) means "he wrote". It consists of two morphemes: the root k-t-b means "write", and the pattern _a_a_a means "3.M.SG.PERF".
I could gloss this following Rule 8:
k-a-t-a-b-a
write-3.M.SG.PERF-write-3.M.SG.PERF-write-3.M.SG.PERF
Or following Rule 9:
k<a>t<a>b<a>
write<3.M.SG.PERF>
Or following the optional Rule 4D:
kataba
3.M.SG.PERF/write
But none of these seems particularly satisfying. What is actually used "in the wild", and what do linguists find most useful when they're trying to actually study a Semitic language?