Re: gdbm support of database.c

Mikio> Diff for ndbm.h is only to avoid confliction between gdbm and dbm Mikio> support. If we choose we can only one of dbm or gdbm (the original Mikio> state of XEmacs 21.2b8), that's not necessary. Mikio> However, I think it's better for users to choose apply patch to a Mikio> part of an operation system. I can open both of dbm and gdbm file Mikio> with such diff. Or should I think to send patch to gdbm support Mikio> group? Changing a file that is part of another package is simply not an option. Even if a new version of gdbm is released with your change applied, people will be using the old one for a long time. I think backward compatibility for stable software must be maintained for 5 years, unless there is a very good reason not to. Mikio> Do people who implement new code want it works on an opetation Mikio> system he/she uses? I use Linux box (Slackware 3.5) which has glibc Mikio> 1 (not includes dbm related function) and GDBM, and DOESN'T HAVE a Mikio> berkeley db. That's a normal PC UNIX user (I don't say not all PC Mikio> UNIX users are same condition). I often write code that only works on operating systems I don't use. It's hard. One of the things that are hard about maintaining XEmacs is the number of choices that users have. You could try installing Berkeley db, and configure with different combinations of database support. Mikio> One day, he/she found that XEmacs could open dbm file directly and Mikio> was glad to open a gdbm file made by another program of which Mikio> extention is neither .dir nor .pag... He/She wondered, why he/she Mikio> could not open the gdbm file even if he/she built XEmacs with the Mikio> --with-database=gnudbm... Mikio> It's right that a problem of maintenance you said exists. But I Mikio> cannot find out why berkeley db should have priority than gdbm... Not priority. Just equal support. Mikio> I'm not sure of text encoding. But I can put Japanese text Mikio> (euc-japan encoding) to gdbm file, and get it properly to other Mikio> encoding such as iso-2022-jp or sjis. But I can get nothing (but Mikio> sometimes get XEmacs core) when gdbm file made by perl. I'm not sure Mikio> that is a problem of XEmacs, perl (Japanized version) or my patch... Don't work on that. It is a difficult job that requires deeper changes to how encodings are handled in XEmacs. It is a job for me. Mikio> You say I should use `||EQ' instead of `|| EQ'? If so I will do. Keep the code in the same style as it is now. Domo arigato. Martin