Hi,
This email has been written quite a few times this morning. You don't
have to read all those versions. That's the first bit of good news.
The second bit is, after a morning and more's investigation, I think
I've found the source of the intermittent test failures I was suffering,
particularly only on a distcheck.
The bad news is it's a grubby little bug which doesn't feel very
rewarding to find, assuming I'm right.
Make the terminal window narrow, say 42 columns. Run ‘make check’.
Observe failures.
It's intermittent for me because I often unmaximise the window running
‘make distcheck’ back to its tiny size whilst I wait for it to finish.
Except when there's failures. Then I leave it maximised to observe a
re-run. Most of the time. Thus, intermittent.
BTW, I noticed a misleading indent going all the way back to 1.1 of
docs/historical/mh-6.8.5/uip/RCS/mhlsbr.c,v:
1179 if (!(c1 -> c_flags & HDROUTPUT) && !(c1 -> c_flags & NOCOMPONENT))
{
1180 if (c1 -> c_flags & UPPERCASE) /* uppercase component
also */
1181 for (cp = (c1 -> c_text ? c1 -> c_text : c1 -> c_name);
*cp; cp++)
1182 if (islower (*cp))
1183 *cp = toupper (*cp);
1184 putstr (c1 -> c_text ? c1 -> c_text : c1 -> c_name);
1185 if (flag != BODYCOMP) {
1186 putstr (": ");
1187 if (!(c1 -> c_flags & SPLIT))
1188 c1 -> c_flags |= HDROUTPUT;
1189
> 1190 cchdr++;
> 1191 if ((count = c1 -> c_cwidth -
> 1192 strlen (c1 -> c_text ? c1 -> c_text : c1 -> c_name) -
> 2) > 0)
> 1193 while (count--)
> 1194 putstr (" ");
1195 }
1196 }
The lines marked ‘>’ should all be indented a notch. I'll make that
change on the assumption they're intended to be non-BODYCOMP lines as
that's how they've been working all this time.
--
Cheers, Ralph.
