This -e and --remove discussion seems to be kind of offtopic here, but anyway, there was a "non-ideal behaviour" I would say that was very well documented https://hashcat.net/trac/ticket/153 and https://hashcat.net/forum/thread-2329.html and was fixed. The problem: by using external salts each hash could be combined w/ each possible salt, this grows exponentially fast and could end up in a very large file. The solution was: not write the salt to disk at all, but only the hashes.
Back to SMF:
I can reproduce this "problem" on cpu hashcat. If you have doubts about the format as always see here: http://hashcat.net/wiki/doku.php?id=example_hashes
It could be that external salt + SMF was not really supported so far or it is just simply broken.
Could you please open a trac ticket for this problem here: https://hashcat.net/trac/ (if possible w/ a testcase and explain what exaclty *is* working and what *not*, e.g. w/o -e it is, w/ -e not etc) and one of the devs will have a look at it soon. Thx
UPDATE: I was able to reproduce it, opened a trac ticket for you (https://hashcat.net/trac/ticket/163 ) and it seems that it works w/ cpu hashcat 0.46 dev version now... Bad news: you need to wait for next release.
Back to SMF:
I can reproduce this "problem" on cpu hashcat. If you have doubts about the format as always see here: http://hashcat.net/wiki/doku.php?id=example_hashes
It could be that external salt + SMF was not really supported so far or it is just simply broken.
Could you please open a trac ticket for this problem here: https://hashcat.net/trac/ (if possible w/ a testcase and explain what exaclty *is* working and what *not*, e.g. w/o -e it is, w/ -e not etc) and one of the devs will have a look at it soon. Thx
UPDATE: I was able to reproduce it, opened a trac ticket for you (https://hashcat.net/trac/ticket/163 ) and it seems that it works w/ cpu hashcat 0.46 dev version now... Bad news: you need to wait for next release.