> Using the COMMENT feature to add guidance to an ATTRIBUTE works great, but
> is it possible to add a TOOLTIP <ATTRIBUTE> = so as to save screen space? 

Where should the tooltip appear? In the entry form or when an entry is
displayed, or even in the tabular listing? I can put a title arround the
attribut names such as

<div title="Please enter some subject here">Subject:</div>

which gets interpreted by most browsers as a tooltip. However it's not
apparent to the user that leaving the cursor on top of "Subject" opens a
tooltip, especially since the cursor is displayed as the text cursor (vertical
bar), not as an arrow over the attribute. Do you have a link to a public web
page which uses tooltips?

> For a complex entry form it is desireable to try and have everything in a
> single window and this, I think, would help.

Aiming for running elog on you PDA? Let me know if you succeed...

> A temporary fix for this is to set summary lines = 0

Right, that's the only way. I would call this "permanent fix" (;-)

Ok, I added the option

Tooltip <attribute> = ...

I apply the HTML "title" tag to the whole table row, so the tooltip appears on the
whole line, not only the attribute name. I guess this is much more intuitive. Give
it a try. New version under CVS and available as a snapshot.

> Ok, I think I see now.  I was assuming it applied to the entry form as well.

Well, good idea. I implemented that. Have a look at this forum (just hit reply to this
entry) and you'll see it. I'm not completely happy with it (at least on small 1024x768
screens), but it's configurable anyhow. New version under CVS.

> Is it possible to add a flag in the configuration file, to specify the 
> maximum size of an attachment?

I implemented a "max content length = xxx" flag, which defaults to 10MB. If an
attachment is larger, an error gets displayed. The problem is that elogd can
only send the error message to the browser *after* it received the whole
request, so it has to "drain" the 240 MB first, which could take quite some
time on a slow connection.

Alternatively, I can cut the connection to the browser immediately, since the
HTTP header contains the content length after the first ~500 bytes, but in
that case the browser only shows a message box "Document contains no data"
which gives the user noe clue of what's wrong.

Anyhow the biggest problem of elog in high stress environments is that it's
single threaded. So if someone issues a search in a large logbook, it could
take some time, in which elog cannot respond to other requests. Maybe I will
find time in the future to make elog multithreaded, but certainly not in the
near future.

> I followed copied the example config provide to run on my server .
> Apparently , the author field was blank when i do a reply .
> I was hoping to get it to be the actual reply author.
> Just like the way the forum is done up.
> Please help ..

This (new) behaviour came from the request that some people id *NOT* want to
have the new author on the reply, but the *ORIGINAL* (they used an attribute
"Origial author" which should not be replaced on a reply). Therefor I added a
new option to resolve that. So please use

Preset Author = $long_name
Preset on reply Author = $long_name

if you want to have the attribut "Author" replaced by the current one both on
new entries and on replies.

> Using "List Display = <attributes>" produces an ususal output with the text
> field content in the first row and a row title of the first 3 letters.

I fixed that problem. Please get the snapshot from July 28th 14:16.

> I noticed that the gcc compiler options in the tarball Makefile were not
> conducive to speed.  There, I tried changing the gcc options to:
> 
> CFLAGS = -O3 -funroll-loops -fomit-frame-pointer -W -Wall

Thank your for this hint, I changed my Makefile for the production code. However, I
could not feel any difference between the two options. The real problem is the
function getcfg(), which gets called many thousand times internally and has to parse
elogd.cfg each time. Once I implement a hash table for that function, elogd should
become faster by at least a factor of two.