Re: [mma] Bug list for mathematica? / Censorship in MathGroup

From: Richard Fateman (rfateman_at_sbcglobal.net)
Date: 06/11/04 

Date: Fri, 11 Jun 2004 14:10:37 GMT

There are good reasons for a vendor not to post a list of
reported bugs, but some do it as a service to their customers
anyway.
If a vendor takes raw bug reports and post them, it presumably
takes some "ownership" of them. Even though 80-90 percent of
user-reported bugs are invalid in some way, e.g.
1. not a bug but a misunderstanding by the user
2. not reproducible (maybe a local problem external to the program)
3. incoherent, incomplete, or imbedded in a mass of irrelevancies
4. not pertinent to the current version of the program.

So vendors don't want to do this. Especially since it exposes
them to legal action ("you knew about this bug"), ridicule in
the market place (X has 200 unfixed bugs...)

Of course WRI has a list of bugs internally. Presumably they
have a checklist that says for each bug if it is fixed or
not in the development version. Unfortunately you don't have
the development version and there is generally no easy way to
give you the fixes that repair your bug on the released version.
And you do NOT want a copy of the development version because it
has other problems. So you get to wait until the next release.

Microsoft has bug lists too, I'm sure. But they don't really want
you to report bugs to them. Instead they suggest you post them to
one of the Microsoft newsgroups, which are UNMODERATED, and which
have participants who are MS employees, also people who appear to
have some substantial expertise but are not employees "MVP"s and
random other people whose questions and advice may be naive, wrong,
or accurate. Presumably bug reports that survive such examination are
picked up by MS>

This is where the mathgroup could serve a purpose by recording
bugs. The moderator might insist that any bug reports
first be sent to WRI, and then perhaps after a suitable short delay
for them to answer, to the newsgroup. In the case of a truly evident
and disasterous bug, it might be published sooner.
Repetitive bugs could be rejected on the grounds that "this is the
same bug as #1234", [probably]. Though that is a hard call.

What about a separate list?

If there were a bug list separate from the newsgroup, it would be
a full-time job to keep it up-to-date. E.g. (broken in versions 2.0,
3.0, ..4.1, fixed in 5.0), or (resolved: user error ) etc.
Presumably this duplicates internal WRI activities.

digression:
One problem with open source in relatively rarified programs like
computer algebra systems (large, complex, requiring advanced knowledge
to write and debug, relatively few users) is that bugs can languish
for a long time; even then they could be picked up by someone who
doesn't really understand the full problem and installs a solution that
is worse than the original problem, just not as obvious.

So is Steve C doing a bad job with mathgroup? It is hard to tell
without knowing what mail he is moderating out of the group. If you
want to become a deputy moderator, say to look over the rejects and try
to get (some of) them published, ask Steve. I don't know how he would
react. I can see pro/con to this.

RJF

Marcus Stollsteimer wrote:
> Hi,
>
> this post did not make it to MathGroup, so I'd like to try here.
>
> My original post (from about 2 weeks ago) was "moderated away",
> oops..., sorry, I mean the moderator took the trouble to answer
> it personally.
> I sent the below reply both to the NG and to the moderator, and
> up to now it neither appeared in MathGroup nor did I get an
> answer from the moderator :(
>
> Regards,
> Marcus
>
> ......................................
>
> Message-ID: <40B70BA7.20306@yahoo.com>
> Date: Fri, 28 May 2004 11:51:35 +0200
> From: Marcus Stollsteimer <marcus314@yahoo.com>
> Newsgroups: comp.soft-sys.math.mathematica
> To: steve@smc.vnet.net
> Subject: Re: Bug list for mathematica?
>
> Dear Steven,
>
> first, thank you for taking the time in replying to my post.
> But I already know that WRI does not maintain such a list,
> I was looking for an 'unofficial' site.
>
> What's wrong with discussing this in the group?
>
> Of course, tech support is very helpful, but from personal
> experience I must say that the effort in exactly locating the
> problem, finding a simple, reproducible example, composing a
> message (I'm a slow writer...), etc., is often a waste of time.
>
> Only to illustrate, here some samples from typical
> replies I got from mma technical support:
>
> - When I run this in 5.0.1 it gives the correct answers. [...]
> you should upgrade to 5.0.1 and see if that fixes things.
> - This is a known problem that has been fixed in the most recent
> release of mathlm (5.0.1.0)
> - This problem has been fixed. You may either download a
> 5.0 binary that fixes the problem here: [...]
> - Our developers are aware of this problem.
> It has been fixed in a future release of Mathematica.
>
> Please don't misunderstand me, I really think the support team
> *is* doing a great job, and the answers are always coming very fast.
>
> And what about the poor users that do not have premier service?
>
> The NG is a terrific resource, too, but it can be really slow
> (e.g., the Eigensystem[] problem needed about 6 days to be resolved).
>
> I just think a list could save a lot of time,
> both for the user *and* for the support team.
>
> Best regards,
> Marcus
>
>
>>Subject: Re: Bug list for mathematica?
>>From: "Steven M. Christensen" <steve@smc.vnet.net>
>>Date: Thu, 27 May 2004 08:09:47 -0400 (EDT)
>>To: marcus314@yahoo.com
>>
>>To: marcus314@yahoo.com
>>CC: steve@smc.vnet.net
>>
>>
>>Over the years, this issue has been brought up many times in
>>various ways. The simple answer is that WRI is not going
>>to publish a bug list. How many commercial software vendors with
>>thousands of customers using many levels of a given piece of
>>software on many different kinds of computers do you know that
>>do this?
>>
>>I work with many hundreds of open source software products that do
>>have open bug lists and those lists are often mostly useless to the
>>general user. Many bug reports are contradictory or just wrong.
>>Even bugs fixes reported by the people who write the code only
>>apply in some cases and fail in others. Some even introduce new
>>bugs.
>>
>>Your best bet is to simply post your bug or question to WRI
>>or if you don't get a reasonable response from them in a given
>>time, post it to the list with a clear statement of what you
>>are doing and what your problem is. Or if you just have doubts
>>about a function or piece of code, send that in. With thousands
>>of users on this list, many of them brilliant expers, you are
>>likely to find a solution to your problems.
>>
>>Steve Christensen
>>Moderator
>>
>>
>>
>>>To: comp-soft-sys-math-mathematica@moderators.isc.org
>>>From: Marcus Stollsteimer <marcus314@yahoo.com>
>>>Subject: Bug list for mathematica?
>>>Date: Thu, 27 May 2004 12:26:30 +0200
>>>
>>>Hello NG,
>>>
>>>this question was triggered again by the recent thread
>>>"Eigensystem: different solution ..."; an earlier query
>>>to support didn't result in a satisfactory answer...
>>>
>>>So: is there a place where I can find a bug list for mathematica,
>>>and (technical) information on new minor releases?
>>>There seems to exist no official information from Wolfram
>>>(see the reply from support below).
>>>
>>>IMO, that's a Big shortcoming on the side of Wolfram.
>>>As far as I understand it, they *try* to inform me about
>>>fixed bugs, but only when *I* previously reported them...
>>>(they don't even maintain a list internally?!?)
>>>
>>>Regards,
>>>Marcus
>>>
>>>
>>>support@wolfram.com wrote (2004/03/30):
>>>
>>>>Hello,
>>>>
>>>>In general as errors of the sort you were experiencing are
>>>>reported to us we send them to the developers and corrections are
>>>>implemented. We do not maintain a list of corrections as they are
>>>>made and so it is not possible to post them where users might
>>>>find them. If you have premier service we try to let you know
>>>>when a new version has been released which corrects any errors
>>>>you have reported, but as of now that is the only mechanism we
>>>>have for informing people of updates.
>>>>
>>>>Sincerely,
>>>>...
>
>

Relevant Pages