home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sources.d
- Path: sparky!uunet!twwells!frnkmth!frnk409!austin
- From: austin@frnk409.franklin.COM (Austin G. Hastings)
- Subject: Re: comp.sources.reviewed and a blast from the past
- In-Reply-To: davet@cbnewsj.cb.att.com's message of Fri, 21 Aug 1992 12:23:54 GMT
- Message-ID: <AUSTIN.92Aug29135259@frnk409.franklin.COM>
- Sender: austin@franklin.com (Austin G. Hastings)
- Reply-To: austin@franklin.com
- Organization: Franklin Electronic Publishers, Inc., Mt. Holly, NJ USA
- References: <1992Aug14.150535.15169@rick.dgbt.doc.ca> <1992Aug17.023423.14527@virtech.uucp>
- <Bt5M67.273.2@cs.cmu.edu> <1992Aug21.122354.12367@cbnewsj.cb.att.com>
- Date: Sat, 29 Aug 1992 17:53:03 GMT
- Lines: 94
-
- In article <1992Aug21.122354.12367@cbnewsj.cb.att.com> davet@cbnewsj.cb.att.com (Dave Tutelman) writes:
-
- >Of course, the whole point of c.s.r is to post only software that meets a
- >standard of quality defined by the reviewers. I'm just wondering if the
- >standard is currently set a little too high, or if there is maybe not enough
- >flexibility in the process to accommodate reviewers' opinions versus
- >authors' opinions.
-
- I think Tom is onto something here. Frankly, quality IS the issue.
-
- When I develop software on the job, my company is going to sell thousands
- of copies of it. Customers have expectations that it will JUST PLAIN WORK.
- If it fails to meet that expectation (1) we spend big bucks on tech support,
- or (2) worse, we lose the customer.
-
- For these reasons, I fully EXPECT my code to undergo:
- - Peer inspections. ANY nitpicky suggestions, even to whether a line
- is adequately commented, is fair game and is usually honored. We
- have a considerable set of methodology and even training in how to
- do code inspections and follow-ups.
- - Extensive testing. A rule of thumb is that the staff-weeks of testing
- a release should approach the staff-weeks of programming it. (And the
- programming includes unit testing anyway.) Before the testing starts,
- there are written test plans (frequently generated with computer aids)
- including test cases and computing environments supported.
-
- As a PROFESSIONAL developer, I actually appreciate this "overhead". It
- prevents quality problems later that are a much bigger headache for both
- me and my employer.
-
- BUT..... I post programs to the net as an AMATEUR developer. The things
- I post are programs I wrote for myself, on my own time. I'm sharing them
- on the net as either:
- - A favor ("I wrote this; you're welcome to use it"), or
- - An ego trip ("Hey, look what I did").
- I'm not doing it to make a buck, and I certainly don't make promises to
- support it. (I frequently DO fix problems, but that ain't in the contract.)
- I believe the great majority of software professionals (that is,
- non-students) on the net operate from similar motivation.
-
-
- Speaking as a consumer, let me interject here that I personally don't care
- what "hat" (professional vs. amateur) a developer was wearing when she wrote
- and posted her code.
-
- There are exactly TWO tests that net.software has to pass in order to be
- used here. In the tradition of classic 'C', failure of the first test
- means that the second won't even be tried:
-
- 1) The software has to work.
-
- It is absolutely astonishing to me how many packages fail this test. Even
- when the packages come from people who CLAIM to be using the same development
- environment that is used here (SunOS 4.x -- not a rare one, I hope!).
-
- It is also kind of amazing that when people complain about this fact, they
- get shouted down. I'm not going to make too much noise on the issue for
- exactly this reason.
-
- Regardless, I feel compelled to reiterate this: If you are a
- net.software.author, and you post your amazing cool software, and some
- net.software.consumer pulls it off the net, and it won't even compile on
- her envionment, .......FLUSH........!
-
-
- 2) The software has to be useful to someone.
-
- This one is like gravity. It's there, you live with it.
-
-
-
-
- IMHO, the noise that's being heard about the quality standards being too
- high may be accurate -- MAY BE. The only test is going to be feedback
- on (and presumably to) the c.s.r group itself.
-
- If the quality control is good enough that people can drop it into place and
- have it work, then it's good enough. The only question remains, is the
- current c.s.r standard higher than even that. (I dunno...)
-
-
-
- Another thing worth mentioning: when in the "plug it in, turn it on" phase
- of a new piece of software, I'd be more willing to believe that some
- failure was my fault if I knew that a "Mongol Horde" of reviewers had
- pounded and abused the software, and they DIDN'T have this problem...
-
-
-
- --
- **************************************************************************
- ** DEBUGGING is no substitute for Austin Hastings **
- * TESTING is no substitute for Austin@Franklin.Com *
- * DESIGN is no substitute for USA+[609] 261-4800 *
-