Dear Mrs. and Mr. Forum,
today Charles Wright sent a bug report to 'gap-trouble' in which
he points out the following.
If you are computing with cosets in factor groups that are not
represented as Ag groups then you might get an error message.
Here is a possibility to produce this.
g:= SL(2,2); h:= SylowSubgroup( g, 3 ); f:= g / h; Cosets( f, f );
As usual, this bug will be fixed in the next version of GAP.