I have just installed 110.43 from ~x86 and ... It compiled and the thing works, but it proudly presents itself as ver.042: wasowski@klimt wasowski $ sml Standard ML of New Jersey v110.42 [FLINT v1.5], October 16, 2002 - If you firstly download the filenames using .42 ebuild (using --fetchonly, or simply by failing to build .42 with gcc 3.3, which is probably what happens to most people), then you have tgz files of .42 in distfiles and the .43 ebuild does not download the new files. The .42 sources are happy with the new patches, so everything goes seemlessly only user gets a wrong version installed. I checked that this is not the bug in smlnj implementation. The files on the official website have the right (.43) version in sources. Then I found that digest-smlnj-110.43 and digest-smlnj-110.42 in portage are identical... No doubt that this will be happy with files from 0.42. Moreover anybody installing .43 a fresh is bound to file on MD5 check [which is a moderately serious problem, as it cannot be solved without hacking in the tree]. The digest has to be recomputed. A side remark: I do not think we should mark this "SML/NJ working versions" stable as they are not considered stable releases by developers of the package. As to my knowledge the last stable release was 110.0.7, not available in portage at all.
*** This bug has been marked as a duplicate of 43018 ***