home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.software-eng:4880 comp.lang.c:18204
- Newsgroups: comp.software-eng,comp.lang.c
- Path: sparky!uunet!cis.ohio-state.edu!zaphod.mps.ohio-state.edu!saimiri.primate.wisc.edu!ames!ncar!csn!raven!rcd
- From: rcd@raven.eklektix.com (Dick Dunn)
- Subject: Re: C Code Layout
- Message-ID: <1992Dec11.165835@eklektix.com>
- Organization: eklektix - Boulder, Colorado
- References: <1992Dec7.164154.6312@b30.ingr.com> <BywqDo.F8F@monster.apd.saic.com> <1992Dec10.191917.18184@yarc.uucp>
- Date: Fri, 11 Dec 1992 16:58:35 GMT
- Lines: 44
-
- scott@yarc.UUCP (Scott Beckstead) writes:
- >tomj@monster.apd.saic.com (Tom Johnson) writes:
- [example of revision author/date/synopsis list
- >>I've always wondered about this sort of thing in a source file.
- >>Is this sort of information considered useful?...
-
- Realize again (as others have noted in passing) that "is this...useful?" is
- a separate question from "should it be in a source file?" Lots of us will
- say it's indispensable, but doesn't belong in the source itself; it belongs
- in the information kept by the control system, where you can get to it if
- you need it but it's not constantly in your face.
-
- Scott then talks about noting each change in the source itself:
- >...Tags were left in the source of the form /*SHB01A.1*/ at each
- >changed location. It was then quite clear when/why the change was made
- >and related this to the top line that had this tag...
-
- This only works for a little while. It doesn't scale to longer-term
- development, because the source code gets so cluttered with these little
- comments that you really can't understand their relationship any more.
-
- >...This technique was MUCH better than
- >trying to diff a file that may have been changed 5 or 6 time before I
- >had to look at it...
-
- Your code-control system ought to be able to hand you a "diff" representing
- the changes between any two given revisions. For the differences between
- two consecutive versions, it's about a wash for comments in-line versus the
- diff. If you're stuck in the (all too common) situation of "it worked in
- version 37; it's broken in 42" you'll most likely want to see the
- accumulation of changes between those versions...including the fact that
- some changes may have been made, then undone. Also, what the code does is
- (at least arguably) more important than what it did--in the sense that the
- contents of the code should be most useful to the person working on it as
- it stands. Change archaeology is important but should take a back seat.
-
- But some of that issue--how to understand the changes--is obviously depen-
- dent on how people work and look at code. My main point is that it doesn't
- extend to long-term development...I've actually watched it fail. Took
- about a year of relatively active development to make source code nearly
- unreadable.
- --
- Dick Dunn rcd@raven.eklektix.com -or- raven!rcd Boulder, Colorado
- ...Straight, but not narrow.
-