jcl yes this is a compromise, if you check smaz_compress it does this kind of checks but only in order to glue together consecutive verbatim outputs into a sing

jcl yes this is a compromise, if you check smaz_compress it does this kind of checks but only in order to glue together consecutive verbatim outputs into a single one (up to 256 bytes), but does not back-check if a single byte interruption is not worth it. I think this can be an option that can be done in a second-pass way if a given #define is active. Thanks for the useful comment.