(regarding links being broken because they seem to be rewritten as
https://www.flyertalk.com/forum/![thread number]!/etc.)
Originally Posted by
IBJoel
Thanks Zorak. We'll keep an eye on it, but I've been unable to replicate the bug and if it's on the order of a few minutes, that's not really a window in which tech could detect, review, and fix the problem. Is there any particular forum you might have noticed it happens more frequently on?
When it happens, it affects links in all forums.
Now, as it happens, this reoccurred for me once in the last 24-48 hours, and again fixed itself within a few minutes. But I tried to do some black box poking around while this was going on, and there may (or may not) be a slight refinement to the above description.
I noticed that the particular link which was broken was a link to another FT thread. Specifically, the link in post #2 on this thread:
UAE inc. DXB / Dubai strict new medicine requirements Oct 2018
Under normal conditions, the forum software replaces that with the actual thread title, so the post text reads as "Already under discussion here Quick advice, please: _New UAE requirement to register prescription drugs online?_" (using _ to denote the link text)
Originally Posted by
UAPremExecflyer
When the bug condition exists, however, it reads as "Already under discussion here Quick advice, please: https://www.flyertalk.com/!threadnumber!/etc."
I certainly understand that if this problem fixes itself within a minute or two, there is no practical way to file a bug report and have a tech look at it exactly while that problem is going on.
However these symptoms seem specific enough that a developer with access to the source code could take a reasonable stab at figuring out the bug just by code inspection.
Even as a developer without access to the source code (unless some form of it is open sourced somewhere, I admit I've never really dug deeper into the origins of FT/IB), I would hazard a guess that:
* there is some internal representation of a FT post where links to other threads are stored by thread number
* at display time the software wants to fetch the actual current title of the referenced thread and render that into the generated HTML
* maybe when this lookup fails or times out or something, we see the buggy behavior