|
|
| Author |
Message |
janc
|
|
HTML in Backtalk Postings
|
Mar 5 15:01 UTC 1999 |
The next version of Backtalk has a new feature that allows users the
options of using HTML in their items and responses.
In an HTML posting you can insert "tags" into the text you type in that
effect the appearance and formatting of items. So if you type
This is a <STRONG>great</STRONG> idea!
then when people read the response through Backtalk they will not see
the "<STRONG>" tags, but they will see the word "great" printed in a
bold font. There are lots of other tags in HTML, some of which will
not work in Backtalk responses (non-functional ones include <BLINK>,
<HR>, <TITLE>, <FRAME>, <INPUT> and such like). The tags that are
allowed let you do various things with fonts, changing colors, sizes and
styles, some formatting stuff, such as automatic generation of numbered
lists and centered text, inclusion of images into responses, tables, and
the creation of clickable links.
Since Picospan users probably don't want to see all sorts of tags like
<FONT SIZE=+3 COLOR=RED>Hello!</FONT> stuff when reading the
conferences, the new Backtalk automatically generates plain text
versions of all HTML postings, and stores them where Picospan will find
them. Obviously Picospan users won't see images or font changes, but
the text, at least, will be legible. Currently the plain-text
conversion isn't amazingly well done. There is room for improvement
here (but it's a fairly complex thing to do well).
I have installed a test version of this new Backtalk as "bt.new" here
on Grex. It has been "broken" so that it only works in the "backtalk"
and "backtalk2" conferences. The following link will run the new
Backtalk in the backtalk conference:
http://www.grex.org/cgi-bin/pw/bt.new/pistachio/confhome?conf=backtalk
Note that if you want to access any other conference after running this,
you'll need to edit the "bt.new" in the URL to be "bt" instead.
The open question is whether we want to enable HTML postings on Grex.
A couple notes:
- There is a switch that will allow fairwitnesses to turn HTML posting
off in their conference.
- Entering HTML postings is actually fairly finicky work, if you want
to do it well. I suspect that they will never become a large
fraction of all postings.
|
| 92 responses total. |
aruba
|
|
response 1 of 92:
|
Mar 5 15:19 UTC 1999 |
If someone enters a response through Backtalk which *doesn't* contain HTML
tags, do you still store it in two different places, Jan? If so, does that
mean that we're essentially doubling the space needed to store conferences?
If those fears are unfounded, I think we should at least try it and see what
happens. Congrats on doing all that work.
|
janc
|
|
response 2 of 92:
|
Mar 5 17:18 UTC 1999 |
There is a pull-down box above the response box from which you can
select the style of response you are makeing, the options being "plain
text", "lazy HTML" or "pure HTML". If you select either of the latter,
then two copies of your response are saved, whether there are tags in it
or not. If you select "plain text" only one copy is saved.
This does waste some disk space. If we could modify PicoSpan, we could
put the HTML to text conversion into it, and only save one copy, but we
can't (and don't really want to). However, bbs disk space is really a
tiny fraction of the disk we use, so I think we can afford this.
|
devnull
|
|
response 3 of 92:
|
Mar 5 19:55 UTC 1999 |
You could change the pager that picospan uses to display things.
A lot of the HTML people might use can probably be rendered on a vt100
(bold can, as can underline).
Of course, we could just ditch picospan, and have the bbs command start
up lynx. I am not in favor of this, however.
|
janc
|
|
response 4 of 92:
|
Mar 5 20:32 UTC 1999 |
No. Lynx sucks in too many ways. For one thing, in the last version I
saw, <TEXTAREA> was implemented very clumsily (no autowrap or scroll),
so that it is kind of a pain to enter responses through lynx. And there
are basic ways in which a command-line interface is nicer than a point
and click interface (especially a point-and-click interface emulated
with character graphics).
I thought about building HTML->text translation into the pager. It has
some advantages over what I am doing here, but it requires having
everyone configure the right pager.
|
janc
|
|
response 5 of 92:
|
Mar 5 20:54 UTC 1999 |
As long as we are discussing Backtalk changes, here's a couple other
things that are likely to change in the future.
- When you join a new conference with Picospan, the first and last
items are "new" while all other items are "unseen". The "new" items
show up as soon immediately, but the "unseen" items only show up
after there has been a new response to them. This is supposed to
prevent you from being hit with hundreds of items on your first
visit.
Currently Backtalk doesn't do this. When you join a conference,
all items are "new". The next version of Backtalk will do the same
thing as Picospan by default, but it will let the fairwitness change
the rules. They'll be able to set specific lists of items to be
new to newusers, or they'll be able to say that all items that have
had activity in the last N days should be new to newusers. This
will only effect people who first join the conference with Backtalk,
not people who first join it with Picospan. After a conference has
been joined, both programs will do the same things.
This will probably be ready within a week or so.
- The next big thing in Backtalk is E-mail notification. I'm not at
all sure we want to enable this on Grex. What you'll be able to
do is flag particular items in conferences as being of special
interest to you. Then you will get E-mail automatically sent to you
notifying you of any new responses in those items. It might be just
a note saying there are new responses, with a URL you can go to to
read them, or it might be the text of the new responses, with a URL
you can go to to respond to the item. You will not be able to
respond to items via E-mail.
I still need to figure out how a lot of that is to be configured.
It is mainly a feature designed to be useful in low-traffic
conferences, which systems like HVCN have a lot of. In such places
people tend to stop coming by to look for new responses because there
are too often none, so slow conferences tend to die. E-mail
notification would let slow conferences maintain activity better,
since any new response will pull people back in, even if they haven't
been around for a while.
There are obvious problems with this on Grex. We don't want large
numbers of users to subscribe to every item on every conference on
the system, because we'd be generating too much E-mail. Some kind
of limits have to be placed on things.
This feature is still just something I'm thinking about. It's
definately going need some discussion of what form we want to
support this in on Grex. My inclination is to enable it to a
significant degree. We've never wanted to be world's greatest
free shell system or world's greatest free E-mail system, but I think
we would like to be world's greatest conferencing system, and I think
this is a feature that would help toward that goal, even if it is
burdensome to support.
|
i
|
|
response 6 of 92:
|
Mar 5 22:32 UTC 1999 |
Note that /bbs has been up past 97% full several times recently. (I
think it's down to 87% now.) What's the time-frame for the new backtalk
going live vs. more drive space going live?
|
krj
|
|
response 7 of 92:
|
Mar 6 06:29 UTC 1999 |
Jan, here's a spinoff of #5. I'd like to be able to keep a "personal
hot list" of conferrence items that I really, really care about --
I could hit Grex knowing that I just had ten minutes to read.
Maybe I would describe this as an "express-checkout" participation file,
and then my normal participation file for when I had lots of time to
spend.
And ideally, stuff I read in "express-checkout" mode would be marked
as read in the regular participation file.
This is just a fantasy, but it would interest *me* a lot more than
e-mail notification. Since I hit grex 5 or more times each day,
I wouldn't need e-mail.
|
steve
|
|
response 8 of 92:
|
Mar 7 03:29 UTC 1999 |
The fluctuations in /bbs are me, not the conferences.
<confessionmode ON> I have spread my corpulance into /bbs since
/a and /c were so full. I have since (mostly) reformed and have
released that space (burp). <confessinmode OFF>
On a more serious note--and why I used /bbs--the conference usage
has got to be the most efficient usage of disk that Grex has. 1M of
disk goes a *long* way. The previous version of Agora is only 4.2M,
for example. Having responses doubled in size doesn't bother me much.
But would everything double? I'm confused on that one.
|
devnull
|
|
response 9 of 92:
|
Mar 7 04:04 UTC 1999 |
It seems like there should be a disk which exists specificly for staff
home directories, perhaps...
|
steve
|
|
response 10 of 92:
|
Mar 7 06:56 UTC 1999 |
Well, yes. I think there will be at some point not so far off.
Whenever I feel human again I intend on working with our 'new' 2G
disks again.
|
janc
|
|
response 11 of 92:
|
Mar 7 13:57 UTC 1999 |
Responses would be doubled only if they were posted via Backtalk as HTML
responses. I believe that would be the minority of responses.
Ken: I've added your suggestion to my Backtalk TODO list (which is huge
so don't hold your breath).
|
krj
|
|
response 12 of 92:
|
Mar 7 19:12 UTC 1999 |
It's not urgent, just a fantasy. Unless it tickles other peoples'
imaginations, I would say, don't bother.
|
i
|
|
response 13 of 92:
|
Mar 7 20:46 UTC 1999 |
Re: #8
No confession needed, steve. I run du regularly & understand why "shove
some stuff into /bbs" is often your best available option. (But if you're
feeling guilty, how about rm'ing or chown cfadm'ing a few of those little
dunno-what-they-are-but-look-like-scrap-to-me root-owned files that are
scattered around /bbs...)
|
remmers
|
|
response 14 of 92:
|
Mar 7 21:18 UTC 1999 |
Getting back to the issue of HTML in conference postings...
I see some benefits to HTML posting in a conferencing environment such
as Grex's:
(1) More attractive, readable text via the use of such typographical
devices as proportional fonts, boldface, italics, underlining.
(2) The ability to display special symbols (such as math symbols) and
characters from non-English alphabets (e.g. German umlauted characters,
French accented characters, and so forth.
(3) The capability of displaying images to clarify or illustrate a
point.
But there are some negatives that concern me:
(1) I presume that HTML responses will be shown in proportional font by
default, and plain-text responses in fixed-pitch font. The effect of
intermixing these in random order - which will inevitably happen since
some people will post using Backtalk and others using Picospan - could
be unattractive and distracting. (I don't really know how bad this will
be until I see it.)
(2) Some people will inevitably insist on using "glitzy" features of
HTML in most or all of their responses - larger fonts, colored fonts,
animated GIF's. Even a little of this would be a perpetual annoyance.
(3) With most browsers, it is very easy to paste a prepared HTML
document into a text entry form. (I did so in a couple of responses in
item 8 of the "backtalk" conference. Check it out, using the new
version of Backtalk.) I worry that this would make Grex an attraction
to spammers.
A worst-case scenario would be that conferences in which HTML posting
is turned on become poster children for the "decline of graphics
standards" (see Larry Kestenbaum's item about this in winter Agora) as
well as dumping grounds for spam.
Now, that's a worst-case scenario. The reality might not be as bad. But
these are some issues to think about.
Have you considered HTML subsets as a configuration option, that is,
allowing certain tags to be selectively turned on or off, at the option
of the fairwitness? Turning off <IMG> would discourage spam, and
turning off <FONT> would control certain other forms of potential
obnoxiousness. On the other hand, there are conferences where you'd
probably want to allow those tags (e.g. "classified").
The negatives I've mentioned might or might not be significant problems
in practice. It depends on user behavior.
|
pfv
|
|
response 15 of 92:
|
Mar 7 21:35 UTC 1999 |
I'd also recommend a "storage format" that is NOT html - so that
shell-users can ignore the "glop" and Backtalk folks get it parsed
properly (it's what I';ve been toying with myself).
|
jep
|
|
response 16 of 92:
|
Mar 8 16:40 UTC 1999 |
Will there be a way for Backtalk users to avoid HTML in user messages?
I'm looking forward to seeing what happens when people start using the
new version of Backtalk, but at the same time, I can imagine problems
with HTML in messages which I or others might want to avoid.
|
remmers
|
|
response 17 of 92:
|
Mar 8 17:56 UTC 1999 |
I suppose it could be a configuration option. That is, the user could
specify that they always want Backtalk to show them the plain-text
version of a response, even if an HTML version exists. Sounds easy to
implement.
Speaking of configuration options, are screen colors configurable by
the user in the new Backtalk? Although I mostly like Backtalk a lot,
I'm getting tired of being forced to read certain conferences in color
combinations that I find unlikeable or even downright fatiguing.
|
janc
|
|
response 18 of 92:
|
Mar 9 04:57 UTC 1999 |
I know what you mean. Coop always causes me eyestrain.
But Backtalk has one basic defect that you have to keep in mind. Every
new page is a completely new run of the program. That means
configuration files can't just be loaded once and saved in memory, they
have to be loaded and parsed again on every hit of every page. The more
configurability I put in, the slower the whole thing gets. Until I
develop a way to save more state between processes (FastCGI or the moral
equivalent), I'm inclined to keep the number of different configuration
options modest.
So I'm not very enthusiastic about letting fairwitnesses ban individual
HTML constructs. For one thing, it's confusing for users to have to
remember which HTML tags work in which conferences.
My notion in designing the Pistachio interface was that fairwitnesses
should be able to give their conferences distinctive looks. I like that
a lot. Conferences with default color settings actually kind of annoy
me. It seems wrong for the staff conference and agora to be the same
color. So, though I find the contrast between text and background too
low in "coop", I do like the fact that I can recognize that it is "coop"
at a glance.
If I let user's override conference background colors, then either all
conferences turn the same colors, which is deathly dull, or I have to
let users configure conference colors on a per conference basis.
Theoretically, the latter solution is fine. In practice, I'd have to
figure out how to store not just setting for each user and settings for
each conference, but settings for each user/conference pair. And I have
to figure out how to find and load those settings fast on every hit by
every user on every conference page. Glug.
Futhermore, this messes badly with people using
<FONT COLOR=RED>Hello</FONT> stuff in HTML postings, because how can I
be sure that no user has a red background?
So for now my solution is that you either (1) complain to the
fairwitness that his conference colors stink, or (2) use your browser's
preference menu to override the conference color, or (3) highlight the
text with your cursor. I use option 3 to read coop.
|
davel
|
|
response 19 of 92:
|
Mar 9 12:02 UTC 1999 |
Some of us *like* deathly dull, Jan.
|
remmers
|
|
response 20 of 92:
|
Mar 9 20:10 UTC 1999 |
I don't get the "deathly dull" bit. I did conferencing for years with a text
interface that always looked exactly the same, and found it continuously
enjoyable. So did you.
Somehow, setting up a system in which one has to worry about what color
someone might be entering their text in seems antithetical to the notion of
conferencing being for good conversation.
|
mwg
|
|
response 21 of 92:
|
Mar 10 20:12 UTC 1999 |
My experience in the old Dial-Up BBS community showed that the ability to
fancy up messages with color and such effects had the fairly consistent
result of reducing the system to a drivel farm. Color became an "avoid
flag" when I was wandering around systems. I'd much rather see the whole
idea dropped like a hot phaser than implemented, because this road has
been trod before, and the serious conferencing types do *NOT* want to go
there.
Responses to the effect that the color codes and such will not be visible
to the Picospan users aside, I expect it to be very easy to identify
heavily-"treated" messages from thier "content".
|
other
|
|
response 22 of 92:
|
Mar 11 02:27 UTC 1999 |
i suggest a counterpoint. making the html tools available may well increase
the amount of junk postings, but they will also offer greatly improved
referencing and illustration of discussion points, especially with the
abilility to paste URLs and tabular information with assurance that the
formatting will remain intact.
|
rtg
|
|
response 23 of 92:
|
Mar 11 09:22 UTC 1999 |
I hate to rain on your parade, but I'd veto all HTML postings, period.
There's so few people who know how to do it well, and so many who do it
poorly or inappropriately, that it gets in the way of understanding
people, rather than facilitating it. If the response box is just a text
box, and there is no editing help to make sure the tags entered are
syntactically correct and balanced, I can predict more eyestrain from
even the unintentional typo.
What I would like to see:
1) repeat the item #0, or at least the title, just above the response
box, as a reminder of what the topic is, and maybe we'd have less drift.
2) Provide a link by each response, with target mailto: the person who
wrote that entry. Hopefully, this would re-direct the chatty asides and
jibes to email, and streamline the conference, so I wouldn't have to
stay up all night reading...
3) Make the system more heirarchical, less linear. I'm thinking like
the old compuserve forums, where it was possible to respond to a
response directly, and the system kept track of the threading. It's
annoying here to read something, and want to respond, then I have to
wait until I've read another forty responses before I get a response
box. By then, I'm sure no one would know what I was trying to say.
|
scott
|
|
response 24 of 92:
|
Mar 11 11:43 UTC 1999 |
I disagree with rtg's point 3 right above... I like the Grex "one long
thread" approach much more than the Usenet model of many forks. I hate having
to navigate all those little forks...
|