In T423528#11827470, @Novem_Linguae wrote:

Same caveat as that task: I understand this is discouraged behaviour now (see T355594: For global elections, stop creating eligible voters table for each election on every wiki and keeping them forever) but there is no alternative at the moment.

@SD0001 wrote a tool to grab voter info without creating SQL tables. T399418: Request creation of voterlists VPS project. Maybe not for this election, but that might be a good tool to try on a small, low risk global election, and then if it goes well, adopt its use more widely. That could begin the process of transitioning off the "create 1000 tables" model of populating global election eligible voters.

Yes - the results are:

Just in case there are still concerns with the above I have set it to start at 06:00 UTC on *Thursday*. We can move it again if necessary before the poll begins, but it is harder to move it after it has already started.

I can set this up (it will be quite tight for Nahid I think if you want to start it this soon).

In T415482#11555028, @Ibrahim.ID wrote:

إسلام

Could I please ask that this conversation move to a more appropriate venue? This task has been declined for a while now.

In T398299#11202639, @Novem_Linguae wrote:

I'd support removal. In another ticket, Dom suggested asking @jrbs if they use the feature. If they say no, then I think that'd be a green light for removal.

In T412254#11448626, @A_smart_kitten wrote:

I may be wrong but I think you might be able to create a Phab project for this tool from within Striker

Yes, but we are getting 15-20 duplicate emails per day, and this is quite a burden on our small support desk.

This is actually handled entirely by triggers apparently :) I have set those up and I believe all is well.

In T355594#11378867, @Ladsgroup wrote:

@jrbs can I drop securepoll_u4c2025_edits everywhere now?

I believe this is resolved with Tran's work above. Thank you so much.

To clarify, this is referring to the Echo notifications and not the emails sent with sendMail.php, which open with Dear $USERNAME, I think. It may require upstream changes to how Echo distributes email since if I remember correctly it is not possible to include variables in these messages.

Yes, that sounds fair. A good thing to clarify for the 2027 cycle as well.

I filed T407526 for this. The issue is that I forgot to remove those who added their names to the list since the first mail. I apologise for that oversight.

I saw this was reopened and I think I know why. The script uses the nomail list when it is compiling the list for the first time. Reminder emails use that same list.

We plan to run a "reminder" notification as well. I'll file a patch just replacing these messages but I'm not sure what the actual best path is for that.

I added this to the voter info blurb:

Indeed I apparently triggered the first log and get a similar error now:

In T400243#11265107, @Dreamy_Jazz wrote:

Is there a sense of how urgent this is to fix?

Is there a button / link to a noJS view? I'm really hesitant to just disable the drag-and-drop for everyone since it is much more intuitive as a voting solution and works for the majority of people.

I only ran into this once, doing it again seems to be fine.

I guess the quickest solution might be to force noJS view on mobile (even on the desktop view) but I have no idea how easy that is.

These are sending now. Thank you!

Woops, I was querying the wrong table.

This might just be me being stupid since JSutherland (WMF) is in there (that's me!) but I definitely have not voted with that account. So possibly my SQL query is incorrect.

Board vote has indeed started! But I think generally we shouldn't push anything while that is open either, in case it causes a regression that impacts the results.

In T406374#11243555, @Novem_Linguae wrote:

To clarify, does this bug happen before or after clicking "Edit Poll"?

After, on the poll creation / edit screen.

In T406375#11242844, @Pppery wrote:

It won't impact the election. The SecurePoll pages are read-only mirrors of what's actually stored in the database, and are not used as a source of truth.

I'm not sure if this would impact the election (the config appears to be saved correctly so it is probably fine?) but would really appreciate PSI have a quick look at this just in case.

Now that it is October my edit count seems to be fixed. Seems that's not true for everyone?

I have basically resolved this with an Apps Script function. We can focus on T399742

This is done!

Hi @Dzahn, sorry for the vagueness of the request.

This is merged now and requests are coming through, so I'm calling this particular task Resolved. There remains some confusion on T390657.

Sorry for the confusion here as a result of T399749. I'm not sure merging the pages makes sense, since the audiences are probably slightly different: The Meta page is for immediate help getting back into an account (and is Wikimedia-specific) while the MW page is more about the extension itself.

Yes, I think that would work!

In T400666#11184393, @Novem_Linguae wrote:

I think that feature ("Dump BLT" for encrypted elections, with encrypted votes in the BLT file) was never coded or is broken. But yeah, if it were working, I think it'd need to collect a decryption key in the UI, then decrypt, in order to be useful. I can't think of any use for an encrypted BLT file. The votes would also need to be shuffled to preserve voter anonymity. (Easy to do, but mentioning so we don't forget.)

In T400666#11169699, @Novem_Linguae wrote:

XML dumps of encrypted elections can be downloaded, but are encrypted [...] Probably wouldn't be useful in solving this ticket unless it were decrypted.

I think this just needs a decision on how to handle this part of the code:

In T404060#11170614, @tstarling wrote:

The election was missing the needs-central-list property, so it was trying to mail all users, not just qualified users. All files generated so far should be deleted. The files should be smaller when the script is run again with the property correctly set.

Got to the bottom of it. All jobs should have been terminated now. There was an issue with the script that was compiling lists with waaaay too many users due to a missed step in the procedure. My apologies for that.

I have deleted all the ml-* files, some of which were incredibly big. So hopefully that resolves the issue.

I'll shut them down. Working with Tim on what happened here

Thinking out loud a bit here, is it possible to convert XML to BLT?

I think we probably should split this into two tasks, one for the design of such a form and one for grabbing requests sent through it and passing them to Zendesk.

The Connector seems to work for this, but it's quite slow. It should suffice for what we want here unless we can leverage the API at the same time (i.e. while solving T404009: Use Zendesk API to get requester emails into the spreadsheet.

The new message will have two targets:

1. https://meta.wikimedia.org/wiki/Help:Account_recovery -> Zendesk form.
2. MediaWiki native form (T399742)

Hi, sorry for my delayed response here.

Just to quickly update here, I have added the voter list to my local deploy1003 space so can run this after voting opens next month.

In T401871#11127749, @Clement_Goubert wrote:

Can you clarify the process? Is one of these scripts using the output file from the other one as input in two different stages, or are the scripts using the output as temporary files in the same run? The first is not possible without workarounds, the second is.

There are two paths forward for this in my eyes:

1. Someone with PHP skills should help rewrite the script to no longer create a bunch of files and instead do both of these stages in memory somehow, if that's even possible.
2. Kubernetes needs to allow us to create and refer to files (writing them to home dir rather than data perhaps?). This is probably much harder and longer-term project.

In T395928#11121946, @Ladsgroup wrote:

Does this make sense? There are three tables that supposed to be empty and droppable but they aren't:

In T395928#11115455, @Novem_Linguae wrote:

FYI, SecurePoll code freeze doesn't end until Sept 10 end of day. T399610: Code Freeze for SecurePoll during Board Elections in September 2025.

In T399610#11117168, @Ladsgroup wrote:

Has this period changed due to the delay of the board election?

Yes, sorry for the delay in updating here. I am anticipating it will move up to match, so will re-begin on 1 September.

My apologies for the delay. This slipped through the cracks of my inbox.