Older blog entries for movement (starting at number 55)

MichaelCrawford : I just marked your "bug report" as INVALID. Why ?

• The report does not target exactly one problem
• The report is vague
• The report does not have a set of criteria by which it can be marked as FIXED, and it cannot be VERIFIED

I would have hoped that you would have read the bug-reporting guidelines first. You just wasted the valuable time of all people Cc:ed on that bug (and the time of another QA to come along and mark INVALID as VERIFIED).

The fact that seems to have passed you by is that everybody has some set of bugs that affect them, and this set differs from person to person. It's a pity your wife feels she cannot use Mozilla; but this is NOT the way to solve that problem. Pay for the bugs important to your wife to be fixed; fix them yourself. I'm surprised I'm having to explain the basics of how free software works to you !

I believe the thing with new accounts + the expert entry form is fixed now in current bugzilla running on bmo.

The simpler helper you describe would be a massive QA problem due to useless reports and duplicate reports. I think the current helper strikes pretty much the right balance. Mozilla wants motivated bug reports from users who don't know much about the browser itself, but are still able to distill a useful report.

I can't comment on your talkback suggestion, since I don't use talkback. It appears to be this bug. If it differs from that, file a new bug on this one issue. Or just use a non-talkback-enabled build.

btw, have you tried k-meleon ?

forrest : no, every HOWTO is not "just the author's opinion". That is discouraged for some obvious reasons; wherever possible, it is much preferred that the HOWTOs stick the salient facts. In particular editorialising is strongly recommended against.

Sure, you'll find sentences like "Personally, I prefer to do it like this, because X and Y" in most HOWTOs. However, well-written HOWTOs do a good job of clearly labelling such comments as opinion, and usually give some sort of rationale, enabling the interested reader to take on board the comment and make up their own minds.

There's a really good article to be written about women in technology (there's already been an album, after all ;) but this isn't it...

I hope my "whining" about the C++ HOWTO is going to end up positively. It is encouraging that there's precedent for the temporary removal of HOWTOs, and I hope that happens soon. MichaelCrawford, whilst I think your article is certainly useful (the return value not requiring a full definition of the class is something that passes many C++ developers by, for one), I wonder if it wouldn't be better where it is already. I'm of the opinion that programming-related documentation isn't really suitable to the LDP; at least, not in HOWTO form. There are two types of HOWTO: the "overview", and the "step-by-step". There are few circumstances where programming topics fit in here.

logic : I agree, that HOWTO simply has no place in TLDP. It is also full of dubious assertions and generalisations.

I'm far more concerned about the really bad HOWTOs that are still being distributed, like the the C++ one (and all the other Al Dev ones, for that matter). The TLDP team are apparently [planning to] review the troublesome HOWTOs, but the really bad ones like that should simply be excised: they have no value whatsoever, and are harming the reputation of the project as a whole. A real pity.

I still haven't found time to finally finish my own HOWTO. I really should.

mbp : you say there is a switch to return memory to the O/S at a watermark. Do you have a reference for this ? I'm dubious, because of the difficulty (and rarity) of being able return X unused pages starting from the current brk-point. alloc/free tends to clump together, but not quite well enough for this to be generally useful...

Of course, decent libc's use mmap(2) for single allocations over a certain size, and these can be returned immediately (and were, last time I checked this in GNU libc).

It seems in some C libraries, you have mapmalloc(3), but it's not clear if the regular malloc(3) can use mmap(2) too.

zwane prodded me into trying out oprofile on my test box with ridiculous interrupt rates. I got up to 350,000 interrupts/second, then the next highest I tried locked the box immediately. How silly.

Time travel. Gibson Research Corp. Same name ... co-incidence ? I think not.

This is what happens when you attempt to make a simple but massively important user interface fix in a project that is used almost entirely by "geeks". It's quite dispiriting to work with such vehement opposition to usability; anyone with an interest in UI design has seen the same attitude from people who haven't thought things through.

Seth Nickell's little article is a cute introduction to why UI really does matter, and matters for everybody. I defy anybody to read that, and still come away with the (horrifically prevalent) attitude that UI design is all about novice users and "dumbing down". Matthew Thomas has a similar article, though unfortunately his examples entirely neglect the "user expectations" side of the argument. It would certainly be nice to see a project throwing out cruft as he desires; however, cruft in user interfaces tends to exist for the same reason it does in source: it is necessary, until such a time it can finally be removed.

googlism : It's amazing how accurate this thing is.

This and this are handy google interfaces, too.

Whereas this and this are just oddly fascinating.

Nearly forgot, song of the day: Morcheeba & Kurt Wagner - What New York Couples Fight About

Looks like Linus is back. I think I'd better wait before spamming him with patches though.

rkarl: yes, it does do that, but the fact there's no separate "word count" sucks rather. It's quite easy to fix (modulo a decent definition of what a word actually is), but you can leverage LyX's rather spangly conversion framework to make a rather hacky solution :

\format wordcount wc "Wordcount" ""
\viewer wordcount "xterm -geometry 30x2 -e less"
\converter text wordcount "wc -w $$i > $$o" ""

SORN: Ghostwriter by RJD2. Lovely bouncy Scruffular stuff.

Neat.

Useless morons.

Bad bad mood.

This makes me weep. Are we still really so backwards ?

It is absolutely ridiculous that this is the level on which such a UI is presented. Can you imagine the average user trying to work out the "kheader" service is ? Why aren't the services listed like :

• FTP server
• WWW server
• Linuxconf remote administration

"Mount Points" ? Are they serious ? And as for "deamons" and "Bugfixes updates" ...

I hope the other modern distributions do a better job than this.