Hi, First, thanks for the tool. It's a useful one.
But here's a quick wish-list, after using the tool in a real use case: 1) a --version switch, to know which SQLite version is statically compiled inside sqldiff 2) in-row difference: Within a row-pair, generate a minimal UPDATE and omits all same-value columns. 3) ability to ignore some columns when computing the diff. 4) ability to use a UNIQUE INDEX as a 3rd alternative to rowid and PRIMARY KEY row-pairing. And here are my rationals for the above. 1) no symbols are exported. I don't remember when I got that tool. Would be nice to have IMHO. 2) I have blobs in my tables, and small diffs in other columns still want to update the large blobs. Makes for a much larger diff than necessary, and also makes it hard to see what really needs changing. 3) 4) these two are linked. I have randomly generated surrogate keys (i.e. Guids used as PRIMARY KEY, for compact FKs of composite NKs), and would like to use the natural key (NK) instead, for row-pairing, and ignore the differences in the PK. These would be IMHO generally useful features, that could benefit many. I might come around to implementing some eventually, but I wouldn't be surprised if someone beat me to it, and I'm not sure Richard would accept patches either. Thanks, --DD PS: Hmmm, I realize that my child table NKs still reference the parent table Guid PKs, so ignoring the PK in favor of the NK / UNIQUE INDEX would work only for "top-level" tables... Bummer. _______________________________________________ sqlite-users mailing list [email protected] http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
