It is frustrating that we found so many problems within days of the release of v5.3.4. It suggests that as a community we are not doing a good job of testing these new releases.
There is no question that the best way to flush out problems with a release is for as many users as possible to make a test upgrade of as many wikis as possible.
So, I would really appreciate as much attention on this as we can manage. Letās try to avoid the ignominy of having to make a further v5.3.6 bug fix releaseā¦
Thank you for addressing these bugs so quickly! Iām afraid I didnāt do my due diligence testing the last prerelease, but this one seems to fix (ETA: does not fix) the only issue Iād encountered with 5.3.4 so far (a non-fatal āScript errorā RSOE when deleting tiddlers in my most highly-customized wikiā¦ possibly related to the $data bug that #8333 resolves? I hadnāt documented it well enough to reportāand in any case, it appears to be fixed now.) I will keep testing!
I too had recently experienced some javascript errors in trying to delete tiddlers, but it didnāt occur to me to connect it with the new release. Will poke around some moreā¦
FWIW, Iāve briefly tested on my complex two main wikis and see no issues.
@jeremyruston , I think we should always include this link in these pre-release announcements (if you have an announcement template, consider revising it):
Apropos this recurring probem with bug-fixes just after the main release, I think it is partly a consequence of the semiannual release cycle that typically introduces so many goodies. Iām happy with this frequency but I wonder if the versioning form is not optimal for it. Maybe the semiannual release should be 5.3.40 and bugfix releases 5.3.41, or 5.3.4a ā 5.3.4b etc?
Or maybe the last segment only could signify the bugfix so we actually do 5.34.1 where 34 is the semiannual release and the 1 is the bugfix/tweaking. So we semiannually celebrate 5.34 and thatās the official number (no need to modify any release banner).
When editing a plugin and mistakenly removing a comma inside the json structure I get an RSOE. Iām curious why this would happen in edit mode? Is it a bug or to be expected?
The only way I can imagine 8333 is to blame is if you have <$data $filter=...> widgets in your wiki. If you donāt then it very likely is a different issue.
Did it happen easily enough that you can test again with 5.3.4 and find a more definitive clue?
Same question. Were you trying the <$data $filter=...> widget?
Yes, it happens every time I delete a tiddler from the story river. Iād thought there might be some connection because your $data widget bug seemed to affect missing tiddlers. But I hadnāt managed to track down the root cause of the error Iām getting, so Iām really just guessing. Iāll clean out as much unrelated content as I can and put a copy up on TiddlyHost in case anyone would like to take a look.
ETA: May take longer than Iād expectedā¦ in the process of removing āunrelatedā plugin content, I also seem to have eliminated the error in my testing wiki. (Perhaps this happened inadvertently while I was testing the prerelease as well? Itās worth noting, I think, that I started encountering the delete error when I upgraded to 5.3.4; Iād been using the same set of plugins without issue in 5.3.3.) Iāll keep looking.
Youāre right - and not an issue resolved by the current prerelease. I seem to have accidentally āremovedā the issue in the lighter version of my wiki I was using for testing. The feature-complete version still has the same error in 5.3.5-pre.
While testing to see which plugins I could safely remove, I discovered that removing TiddlyMap also seems to āsolveā the problem. I hadnāt had any issues with it prior to the 5.3.4 upgrade, and had not made any changes to TiddlyMap or my page layout since upgrading.
This error is thrown in title: $:/plugins/felixhayashi/tiddlymap/js/AbstractRefEdgeTypeSubscriber which is an uglified library. Libraries like this show up as 1 line in the debugger.
So there is no easy way to find out what really causes the problem. But itās the plugin.
It would be helpful if you could test TiddlyMap with an otherwise uncustomized empty.html and see if the error still occurs. I am not familiar enough with TiddlyMap to be able to do that in a meaningful manner.
I did some more test and TiddlyMap does not work since v5.3.1, where it started to show initialization problems.
TW v5.3.4 loads it, but the library accesses a variable āpā that does not exist.
I did compile the TiddlyMap plugin from source and it seems to fix the ādelete tiddlerā problem. @etardiff Iāll send you a private post with JSON attached for testing.
But I do not know if TiddlyMap itself still works, since I do not know / use it.
I do see the logic; the most recent 5 releases have all required a bug fix release. There are counterarguments, though. Users might come away with the impression that a v5.x.0 release would necessarily be unreliable, making them inclined to wait for the v5.x.1 release. That runs counter to our goal of getting as much battle testing done as possible for each new release.
Thank you, I hadnāt come across that problem. Itās a long standing issue, but the fix is very straightforward so I am merging it for v5.3.5:
Thanks @pmario. It would be good to get TiddlyMap working again.
I think tiddlymap issue must be brought to the notice of @Flibbles since he is trying to maintain that project. I have also noticed this error while trying to delete a tiddler, but only after upgrading to 5.3.4
Thank you for testing and compiling this solution, @pmario! It does seem to resolve the error in 5.3.4 and 5.3.5-pre, and TiddlyMap still works for me.
I hadnāt noticed any issues with TiddlyMap in 5.3.3(-), but itās very possible it was failing in some silent way that didnāt trigger an error message and didnāt impede my usage.
I really appreciate the diagnostic help! @Springer, is it possible that youāre also using TiddlyMap, or should we keep looking for a different issue?
I can confirm itās not TiddlyMap in my case. Alas, I failed to pay attention enough to noting the conditions so that I could revisit the problem, and I havenāt come across the RSOE again since this thread has been on my radar. So, Iām afraid I have to let the conversation move forward without more troubleshooting help from me.
Sorry, I know this is late; a long weekend intervened.
After fighting GigantiCorpās latest productivity-blockers, I was able to test most of my wikis. I found no regressions whatsoever. But I havenāt paid close attention to the bugs from 5.3.4, so I canāt really tell you if it fixes anything.
