updated forum

[sam_]

A forums administrator is now a member of the infra team.

[Goverp]

It should be possible to edit the frontend/HTML code of the forums if your main issue with the update is the looks and layout, shouldn't it?

Isn't that what CSS is all about?

[Zucca]

Isn't that what CSS is all about?

I've been using my custom CSS fro fgo for a while now.

[b11n]

It should be possible to edit the frontend/HTML code of the forums if your main issue with the update is the looks and layout, shouldn't it?

Isn't that what CSS is all about?

only if the HTML is generated with CSS in mind (which everybody has been doing since 1998, except phpBB 2.0)

If you generate it as a table, the browser is obligated to render it as a table. No matter the size a a viewport, all the cells in a row of a table sit beside each other horizontally.

[Chiitoo]

Moved from Gentoo Chat to Gentoo Forums Feedback.

As someone who has investigated a forum upgrade years ago, I can say the process is not easy. Others have gotten further than I have apparently.

Things to consider:

• There are over 1.1 million threads and nearly 9 million messages if importing them is a priority.
• Some of the old messages include text that look like tags which phpBB tries to interpret but fails and halts the process
• When I tried it, a step timed out and that means you need to start over from scratch as there is no way to resume

Indeed...

Code: Select all

You may only use fonts up to size 200.
In /var/www/forumstest.gentoo.org/htdocs/install/convertors/functions_phpbb20.php on line 1273

Yeah, it does that a lot on like, 'lspci' or 'dmesg' or such output.

I don't know if it changed at some point, but at least since phpbb-3.0.14 it does continue after such errors.

Also, there /is/ a resume option, but it does still start from the beginning for me. Did not really look into it, as I have always been wanting to try a clean run from the start anyblue, but while hunting for certain errors, I did find that I can use the URL to set the starting point, or, resume points manually. Unsure how cleanly it actually works, but at least in theory, if set before the failing point, the SQL queries will just error out due to "duplicate entry", and it's not fatal to the convertooor.

Indeed, I haven't seen any error halt the process aside from one, where it takes some posts with non-UTF-8 bytes, thinks it's all good UTF-8, and passes it on to a text parser where an actual validity check exists... and it fails, hard. Not handled at all by the convertooor code.

Okay I have been seeing a lot of internal server errors as of late, but I suspect that's just our resident crawlers having a party.

The UTF-8 convert-issue I think is one of the main frustration points to the one working on it a few(?) years ago, though not a blocker since it was stepped-over by dropping any such posts (now I know this problem extends to usernames and dateformats as well, and really, anything else too).

Big thanks in any case to antarus for their work on this, and everything else, too.

Think we got close that time, but retiring from Gentoo Linux became a reality first.

I've made some changes to the phpbb convertooor that takee into consideration these encoding quirks, and I dare say things look pretty good.

For the time-outs, we have some pretty powerful infra these days, so can crank things up pretty good so time-outs are no issue from what I can tell.

In fact, right now the most issues come from the mentioned crawliers causing 502s and such... we are working on that too on the infra side, though, so it will get better soon(tm).

Does it take an unreasonably long time before timing out? Even on current hardware? Couldn't the discontinuation of OTW have reduced the volume a little bit?

Besides, is it possible to tell what old threads have misparsable text? Couldn't it be as simple as scanning the db in search of specific text patterns and replace them? Or is it something that can be detected only after conversion is done?

Things are inserted in batches, so it's not a "one big time-out".

~60 seconds usually for one batch, even if I really crank the settings up. I can't seem to make it fail with the memory and other resources we have available no matter how high I do crank them.

We might have that time-out set to one week now. :]

Quote from another thread :

While the forums project members have gone through various test runs of migrating to phpBB-3 since 2017, we still lack the home for it.

We do not have access to the infra knobs that are required, so in addition to our team, we need someone from the infra project to have the time and interest to set things up on their side (the last person to work with us for this retired from Gentoo development).

All we can do at this time, is pester the infra members from time to time to see if we could some day or/and night work on it.

Perhaps 2025 will be the Year of the Gentoo Big Forum Upgrade. :]

Does this still stands true ?

I'd now say 2026 is the Year of the Gentoo Big Forum Upgrade, unless something extraordinary silly happens...

[pa4wdh]

Given the troubles with the update and the growing complexity due to the growing difference between our version and the current release, wouldn't it make sense to set the current forums to read-only as an archive and start a new one with current software?
To facilitate any ongoing discussions you could create topics on the new forums with a link to the old topic (and maybe the other way around) for topics that were active in the last x days.

[Chiitoo]

Given the troubles with the update and the growing complexity due to the growing difference between our version and the current release, wouldn't it make sense to set the current forums to read-only as an archive and start a new one with current software?
To facilitate any ongoing discussions you could create topics on the new forums with a link to the old topic (and maybe the other way around) for topics that were active in the last x days.

I believe it is very doable to preserve the history.

Whether or not it makes sense or is worth it, for actual human users, might be debatable, but I think we're doing it in any case.