recoll review request help : using modified versions of libraries for build

[Jean-Francois Dockes]

Rahul Sundaram writes:
 > I haven't looked closely at this specific case but the general idea is
 > that if a project A is modifying B and not discussing those
 > modifications at all with B, then we end up having duplicated code
 > instead of having project B being enhanced by A's contributions.  In
 > that sense,  it is always a good idea for A to talk to B and see if the
 > changes can somehow be merged back instead of being specialized to A. 
 > If it is indeed impossible, you might ask the person submitting the
 > review to file a ticket with FESCo, make their case and get a exception
 > to our guidelines. 

Hello, this is the Recoll developer here. General rules are good, but
specifics are sometimes interesting too. Here they come:

- Unac: last release: Oct 2004. I have talked with the developer a few
  years ago, about a bug fix, he was not interested in supporting Unac any
  more, and asked if I would pick up the baby. I declined, because I am
  willing to support and extend the Unac package as used by Recoll, not as
  a general purpose library. I have since extended the Unac functionality
  (to case-folding), but I am not willing to publish this as a general
  purpose feature as it was quite closely tailored to Recoll needs.

- Binc Imap: last release july 2005. The site front page states that
  development is halted and no new releases will appear. Binc imap never
  exported libraries anyway. I picked out the MIME analysis part of the
  code because I thought it would be more robust than anything I could
  write, being proven in the field, inside a server. I found a few very
  minor issues, that I fixed, but I don't see them being merged back, given
  that Binc Imap development has been totally halted for the last 5 years.

Especially in the case of Binc, I find it a bit strange that Fedora rules
would appear to forbid code reuse. This goes quite a bit against common
wisdom. Not all interesting code is published as a library, or maintained,
and sometimes the only way to reuse is to copy and modify and it's a quite
valuable way (compared to rewriting from scratch). 

Regards,
JF