The only way I could get it to work properly on macOS is by replacing the hopelessly out of date system TK bundle with something up to date, such as ActiveTK (which I believe is still free for development use), then downloading the UNIX source for Potato (not the Mac binary) and running the script manually. It works but it's a PITA. You'll probably be better off switching to Atlantis.
Sometimes I read on the phone, sometimes on the computer, and for a while, each would track the last post I read on them... so if I only read to page 5 on the phone, read to page 10 on the computer, then went back to phone, it would dump me to page 5.
It stopped at some point, and I think it started again, I'll experiment more.
The 'do the secure installation' step that is now recommended does something that changes the standard permissions that tinymux expects when it comes time to configure it.
I know I managed to fix it once. I do not for the life of me remember how, but it wasn't any of the things recommended on the thread to fix the mysql.h issue, because none of those worked for me at all.
It may have been some kind of worst practice hail mary I yanked out of my ass, so I don't recall, unfortunately, what it was. It was a permissions issue, that much I do remember. Whether it involved changing the permissions on the directory where the file was housed or the file itself, and what I ended up changing them to, unfortunately, I do not remember. 😕
I know that isn't much help, but maybe with that information some of the people who actually know what they're doing can help narrow it down.
ETA: For the tech people: If this is a digitalocean install, mysql.h is there, and it is installed. The permissions are just locked down in such a way that tinymux can't access it to use it to do anything -- which may be only after the 'mysql_secure_installation' step, but even if you did this after setting up tinymux you'd run into the same issue later -- so tinymux acts like it isn't there because it can't see or use it.
ETA2: Mediawiki 1.30.0, the current version, is fussy and different in some new and interesting ways. Be advised if you're using code from an earlier iteration or planning to import. Apparently a lot of stuff deprecated between versions 29 and 30 and you will probably need to tweak your stuff. Depending on how much you fuss with things, this may be no big deal.
(I know I found the stuff they're starting to add in so egregiously annoying I gave up on using Vector entirely and started building a new skin during the time I can't really do anything else productive (while this comp is on its last teetering legs and we settle in for the long wait for its replacement).)
I went into the 'site-packages' directory for Python3 and copied PyQt5/ to Python2's. Hey, at least it's now failing creatively:
$ python main.py
Traceback (most recent call last):
File "main.py", line 8, in <module>
from PyQt5 import QtCore, QtGui
ImportError: dlopen(/usr/local/lib/python2.7/site-packages/PyQt5/QtCore.so, 2): Symbol not found: _PyBytes_AsString
Referenced from: /usr/local/lib/python2.7/site-packages/PyQt5/QtCore.so
Expected in: flat namespace
I have tried and failed to install PyQt4 from the official site, as it complains that it can't find qmake, and I can't find an install for qmake except via the qt-devel library. Until I become a more capable developer, I will now leave this in the hands of other people, and let this thread get back on track.
For those now afraid to try out Digger, I am only having these issues on macOS. Try it out!
Hey, buddy! Welcome to the thread!
This is a followup on my other thread here: http://musoapbox.net/topic/1546/aws-amazon-web-service-as-mu-hosting/
People had expressed interest there, and in PMs, and in chats elsewhere, about what I did to get a MUSH up on AWS Free Tier. I had told them that I would write a how-to... But as said, the guidance I could give was insufficient to mitigate the potential risks of a mistake.
@Cheesegrater above has a good breakdown of how to avoid those problems. It's much less dire, given that info... but given all the other options at comparable cost, it's just not worth the potential trouble to me. If/when I go to Linode or whatever, I'll write that how-to instead.
Maybe I'm crazy. I probably am. (Okay, definitely.) But I'm curious.. how difficult would it be to configure the list of displayed jobs (i.e. when you type +jobs) to include the name of the submitting player as well as bucket, job name, due date, assigned, status?
(Ssssh, I'm not supposed to post.)
But this is actually already a function. You just change your &jobs me=<stuff>. For example: &jobs me=BOTD will give you bucket, opened by, title, and date. &jobs me=BOTDS gives bucket, opened by, title, date, status.
In fairness, the suggestion I have for someone to be able to common sense fix the problem the current mediawiki install has on digital ocean is not a best practice.
I am reasonably sure there's a more secure way to do it than the one I laid out through some back end, and the one provided does absolutely have risks (and big ones).
For someone without heavy duty SQL/mediawiki/server experience, however, it works and will fix the problem without incident most of the time. That describes me, really. I only know the fix because I ran into the same problem, and after digging around about how to go about it, that was the only answer I found.
At some point, you have to do what works and hope for the best, especially when the argument about how to get something done takes up more time than researching how and getting it done.