Just spotted a bug in forum_search_action.php

Doing an advanced search (here, at SETI, or at SETI Beta), you get a list of messages back.

At the end of the header row for each message, there's a forum link to the Message ID - except it isn't. The number shown, and linked as the #message, is actually the thread ID number instead (repeated): a search for 'bug' in this forum returns (inter alia) http://boinc.berkeley.edu/dev/forum_thread.php?id=2369&nowrap=true#2369 - my opening post in that thread was #14479.

Should be easy just to switch the field displayed and referenced in the link.

Sorry Richard, but I can't reproduce this here.

When doing a search on bug, I get lots of threads and those with a message ID number on the end, show the message number, they also point to that message number and when clicked, go to that message number.

I'll test on Seti. Done testing it on Seti, can't reproduce it there either.

So, check if it isn't something local. Noscript blocking Javascript, or cookies not being accepted?

---

Ah - just tried it here, seems to be dependent on whether you restrict the search to a single forum.

Try it again with "Forum
Only display posts from this forum" set to something other than 'All'.

Just tested CPDN, which has the older-style search results format ("[Read the rest of this post]" after each message, rather than an ID number in the message header). No problem there, so the issue may have arisen when they re-designed the results page.

OK, gotcha. Added as ticket [trac]#539[/trac].

---

Ticket has been closed, and the fix seems to have been deployed to the forums I originally reported (here, SETI, SETI Beta).

Thanks, David.

---

More jiggery-pokery with advanced search:

Fatal error: Call to a member function setHighlightTerms() on a non-object in /disks/setifiler1/home/boincadm/projects/beta/html/user/forum_search_action.php on line 169

I'd used searching successfully a few minutes earlier, but noticed another bug and was trying to recreate the circumstances! The problem I saw was with the message "Posted [x length of time] ago by...." in the message headers: under some circumstances (tbc), the [x length of time] seems to be the user's most recent post in the thread, rather than the age of the actual message.

The fatal error has gone away, so I've been able to reproduce the bug.

Again, it only shows if you restrict the search to a single forum thread. If you search on 'bug' in this forum (Web interfaces), it says that Jord's message #14835 was posted 2 days ago. Seems to show all messages found in a thread as if they were posted at the time of the most recent post in that thread.

Edit - and to confirm the point, it's now telling me that Jord's message was posted 17 seconds ago.

Ticket reopened. Add to it if you can. (You do need to be registered there).

---

Fixed again: [trac]changeset:14603[/trac].

I like the description. :-)

---

Fixed again: [trac]changeset:14603[/trac].

I like the description. :-)

Frankly, the busier I can keep him with 'extremely minor bug fixes', the less time he'll have to b****r up the rest of BOINC.

Anyone got a libcurl proxy we can bother him with?

---

I shall download the newest version of BOINC and will probably find that the nomenclature changes I suggested haven't been implemented consistently. So I'll be able to reopen the BOINC vocabulary ticket which I'm sure is considered even more minor by the developers. I'm certain to find extra incomprehensible phrases to complain about, though of course I can't ticket anything until I discover what it means myself.

Don't need for a crowbar to get my jaws to open on the topic on proxies! The one show-stopper for many with smaller and larger farms available to crunch.

Why don't you complain about that on libcurl mailing lists, which is where the real problem is?

---

Found yet another bug in Forum Advanced Search. This time, it happens whether or not you restrict the search to a single message board.

Trying to deal with a bug at SETI which causes major loss of completed work with the words "[Error] Error on file upload: no command" in the BOINC message log. So I did an advanced search on "Error on file upload".

This found several message log excerpts, including some which displayed like this:

ont color="blue">1/25/2008 11:54:14 PM|SETI@home|Computation for task 28no06ah.19864.24612.15.6.131_0 finishedont>
1/25/2008 11:54:14 PM|SETI@home|Starting 28no06ah.30219.5798.16.6.159_0
1/25/2008 11:54:14 PM|SETI@home|Starting task 28no06ah.30219.5798.16.6.159_0 using setiathome_enhanced version 528
ont color="red">1/25/2008 11:54:14 PM|SETI@home|Sending scheduler request: To fetch workont>
1/25/2008 11:54:14 PM|SETI@home|Requesting 14764 seconds of new work
ont color="green">1/25/2008 11:54:16 PM|SETI@home|[file_xfer] Started upload of file 28no06ah.19864.24612.15.6.131_0_0
1/25/2008 11:54:18 PM|SETI@home|[file_xfer] Finished upload of file 28no06ah.19864.24612.15.6.131_0_0ont>

The original message was:

[color=blue]1/25/2008 11:54:14 PM|SETI@home|Computation for task 28no06ah.19864.24612.15.6.131_0 finished[/color]
1/25/2008 11:54:14 PM|SETI@home|Starting 28no06ah.30219.5798.16.6.159_0
1/25/2008 11:54:14 PM|SETI@home|Starting task 28no06ah.30219.5798.16.6.159_0 using setiathome_enhanced version 528
[color=red]1/25/2008 11:54:14 PM|SETI@home|Sending scheduler request: To fetch work[/color]
1/25/2008 11:54:14 PM|SETI@home|Requesting 14764 seconds of new work
[color=green]1/25/2008 11:54:16 PM|SETI@home|[file_xfer] Started upload of file 28no06ah.19864.24612.15.6.131_0_0
1/25/2008 11:54:18 PM|SETI@home|[file_xfer] Finished upload of file 28no06ah.19864.24612.15.6.131_0_0[/color]

Looking at the page source for the browser display for the Forum Search Results, the first line of the examples above looks like this:

<f<span class="highlight">on</span>t color="blue">1/25/2008 11:54:14 PM|SETI@home|Computati<span class="highlight">on</span>
 for task 28no06ah.19864.24612.15.6.131_0 finished</f<span class="highlight">on</span>t><br />

- in other words, the "funky search result formatting" in [trac]changeset:14267[/trac] has been applied to absolutely everything, including search terms embedded in HTML tags arising from the conversion of BBCode 'color' tags.

Not good.

Edit: Trac [trac]#557[/trac]

Thanks for your report about the search system highlighting. The fix for this should be here shortly.

Wow!