List:MySQL++« Previous MessageNext Message »
From:Andrew Sayers Date:May 18 2007 10:45am
Subject:Re: Bug: program dies with untrue error message for template query
View as plain text  
Thanks for the quick response.  This post probably breaks threading - I didn't
bother subscribing until this morning because I didn't expect I'd reply.  I
consider the code bug closed, but here's a suggested patch for the
documentation (to go on the website and in the README):

(Begin quote)

<h1>Bug Reports</h1>

Please report bugs to the mysql++ mailing list: plusplus@stripped.
You can view replies (along with all the other MySQL++ conversations) at or subscribe through the form on that page.

When reporting bugs, please include a sample program that triggers the bug,
along with the expected and observed output of the program.  If possible,
modify one of the MySQL++ examples to give the same effect.  Otherwise, either
use the mysql_cpp_data database from the examples, or include information about
how to build a database that triggers the bug.

<h1>Known Bugs</h1>

Template queries with a single parameter have a number of bugs.  The next
release of MySQL++ will be aimed at fixing these.  In the mean time, you can
use "%1" instead of "%0" in single-parameter queries - you'll have to give a
default value for %0 if you do that, or you can include an empty 0th parameter
whenever you run the query.

(End quote)

Here's my rationale for this patch, if you're interested:

- On rereading it, the website does say "questions should go to the list" and
  "patches should go to the list", but not "bug reports should go to the list".
  Since I was looking for a statement about how to report bugs, those sentences
  only registered faintly in the back of my mind, covered by the louder thought
  "if I'm wrong about that, they'll shout at me for sending a report to the
  wrong place, and I'll look stupid in public".  Maybe I should have been smart
  enough to read both statements and interpolate between the two, but my bet is
  I'm not the only one, and that most people just keep quiet rather than
  risking embarrassment.

- This text uses the words "bug report" and "report bugs" in capital and
  lowercase.  Most people wanting to make bug reports will search for some
  permutation of these words, and when searching the README file, might well do
  so in `less`, which does searches case-sensitively by default.

- It mentions modifying the examples.  I'm used to the general rule being
  "provide the smallest possible code block that triggers the bug" rather than
  "provide a code block that's known to be good except the line(s) that are
  altered".  There's nothing wrong with the latter rule, but you need to tell
  people about it if you want them to follow it :)

- It doesn't mention searching the list archives for bugs, because anybody with
  the energy to search a general mailing list for their specific bug would have
  already done so when they saw the archive, and everyone else would just be
  put off by it (looking stupid in public again).

- Even if people won't trawl ML archives, you can save the bother of discussing
  the most prevalent bugs on a regular basis, while providing information even
  to the people that don't bother writing reports.

Finally, I don't suppose there's a setting (or patch) for ezmlm where, if I send
a post to the list, I can get replies just to that specific thread, without
having to subscribe?  Clogging up an inbox just to get a "this is fixed in the
nightly builds" message isn't something most people will do.

	- Andrew
Bug: program dies with untrue error message for template queryAndrew Sayers17 May
  • Re: Bug: program dies with untrue error message for template queryWarren Young17 May
Re: Bug: program dies with untrue error message for template queryAndrew Sayers18 May