Wiki Home

Wiki Bug Fix Archive

Namespace: WIN_COM_API
..[Title] "No output" Response After Creating New Topic Via URL[/Title]
..[Party type="discoverer"] Randy Pearson [/Party]
]]1. Try to create a new topic via the URL by just typing the new topic name in the URL. 2. You get the (expected) page to describe the topic. 3. Fill out the page and submit. 4. You get a Web Connection response that 'no output was returned'. 5. The description you entered was in fact saved correctly. 6. If it makes a difference, I also included a Category keyword in my description.
Randy, I just tried this and got the same error ('no output was returned') but then I hit the back key and saved again and the save worked. I've been seeing quite a few 'no output was returned' messages, but hitting Back and then save again always works. I wish I knew why that message is sporadically coming up... -- Steven Black
Steve, Actually it was saved the first time, so the bug comes after the save operation. (I determined this by going straight to Recent Changes and refreshing--my topic was there and the page had the new info.) My Web Connection notes on sources of "no output" problems like this indicate a common explanation is in a llNoOutput flag being reversed in one of your wwHTML method calls. A second possibility is some function sending a non-character type to Send() or Sendln(), which don't do error/type checking. -- Randy Pearson
Thanks Randy! You're a pill. Fixed! -- Steve
New->PRE tags are sacrosanct. We should not parse anything in there. See the topic VFPandWindows, for example. It's Double spaced when it should be single spaced, and we shouldn't be wikilinking program code...
(eg) I've placed a fix to take care of the wikilinking but can't figure out the double-spacing. Did you get the example from Fox Help?
No. The problem seems to be that BR tags are being found within the Pre tags. Within Pre tags, a carriage return is sufficient. I think we need to change the CR parser to ignore PRE regions.
(eg) Done. Working but may need rewrite as we are knocking off a cr+lf at the end of each parser execution (less code) rather than making sure the last cr+lf is not added to begin with.
Make the Alphabetic Topic List sort order case insensitive.
"?" preceding word forces a link. e.g. INTLLinks will create a new link.
Can we have the edit page say "Editing [Topicname]" in the header instead of just [Topicname]?
(eg) Done.
Note the ProblemTrailingParenthesis) and the ProblemTrailingbrackets] -- they get included in the link. Note that exclamation points do so too, as in Web Talk! Click these to see what I mean -- you'll get topics with these punctuation suffixes.
(eg) Done.
Somehow we're getting leading spaces in the page source. This should take 2 minutes to fix.
(eg) Done.
Let's make all external non hyperlinks open in a new window.
(eg) Done.
The link for a non-existant topic should be a question mark, or whatever, so the user isn't unwittingly led to empty topics.
(eg) Done. Exclamation mark needs including in punctuation string.
New->ER: Can we explicitly set the background color of the page to white? Otherwise the globe.gif is set in a white square. OTOH we could somehow make the white bits transparent? I know SFA about gifs....
(eg) Problems with MS Image Composer. Will find work around.
[smb]Let's make the page background property white....
(eg) Done.
Let's make all external non hyperlinks open in a new window.
(eg) Done.
Currently a Save inserts a backup record with a Date Time() in TOPICSBAK.DBF, and then it updates or inserts into TOPICS.DBF with another Date Time() which is milliseconds later than the one in TOPICSBAK.DBF. I'd like to see the same timestamp in both the TOPICSBAK.TCHANGED and the TOPICS.TUPDATED fields.
(eg) Fixed.
Bug: A .TopicCookie prefixed by a period doesn't get annointed as a link.
[EG] That's because the rule is that the first letter must be a capital letter. Do you want this changed?
[smb] No, but the first letter is a capital, yes? :-)
[eg] OK, the first character then.
[smb] The first character of TopicCookie is T. That there is no space between it and the period prior to it does not change that.
Bug: "?" and ";" still part of links.
(eg) Fixed as part of your changes, Steve.
Bug: Bolded Email andISBN links are not working as in or ISBN 123456789X.
(eg) Fixed.
Bug: if "mailto:" is entered manually, it appears twice in the link. e.g.
(eg) Fixed.
Go through webtalk.prg and htmlparser.prg and eliminate spurrious LOCALs, and make sure all used memvars are declared LOCAL. Also in htmlparser.prg there are unused methods at the bottom of the file, and the STRTRANC() method needs to be cleaned up.
(eg) Done
In HTMLPARSER.PRG there are still one or two calls to REMOVEPUNC() which should be removed and coded like the other classes in HTMLPARSER.PRG
[eg] Done...
Bug: A BoldedTopicCookie doesn't show up as a link.
(eg) Fixed.
The Find results, from the bottom of the page, should be a single spaced list. While we are there, in the find results, let's also display the date last changed and the IP address of whoever made the change.
I think we should we hook the page production process! In other words, consider where currently the header is reckoned and written to the page. This should be hooked. Same for the footer where Edit/Find are. Call me about this.
Let's put the Recent Changes and StratingPlaces centered at the bottom of the page. Can we extract these links from the PrivateTopic called Footer Links? This would be so cool.
The NHits field in topictext isn't being updated.
Bug: This is a URL entered manually as HTML: Go To CNN. Note how the parser totally mangles it. It shouldnt.
Bug: A .TopicCookie prefixed by a period doesn't get annointed as a link.
Let's make a change to the structure of the backup table such that we have one record per newly saved edit.
The UpdateOn field in TopicText should be type Date Time.
Let's code-review the page generation code, which I see is already getting slower for larger pages like this one.
The auto-generated Recent Changes should be a single-spaced list. Recent changes should be sorted by Date Time descending before display.
RemovePunc suppressed in favor of situation-specific parsing.
I really want topic records to be first created when a user first saves it with content, and not simply because someone saves text with a PotentialNewTopic.
Bug: TopicCookieWithComma, the comma is part of the link topic.
(Fixed. (EG) Ending punctuation is now addressed by a simple call to RemovePunc() throughout Webtalk.
Bug: Find topic titles which did not exist results in an additional header in the FindResults page.
((EG) Fixed..
Buglet: At the bottom of every page, between the last horizontal
line and the Edit/Find links, there is a spurious P tag.
((EG) Fixed..
Bug, see what happens when I type five dashes as above.?
((EG) Fixed but we need to talk about this one..
Are these horizontal lines stuffing an unnecessary carriage return after themselves? It looks like it.
(EG) Fixed..
Bug: This is a URL entered manually as HTML: Go To CNN. Note how the parser totally mangles it. It shouldn't.
(EG) Fixed as an intermediate change. Need to make this a prehook sooner rather than later. Same solution for < pre > tags.
Two Bugs: See what happens if we do this: ISBN 1234567 is the first word missing in this sentence (the first ocurrence of the word "Why' doesn't appear.
Also, why is there a line break in the display after the ISBN?
[eg] Fixed...
Category Wiki Admin
( Topic last updated: 1999.06.14 12:09:51 PM )