@Prtksxna You can use this task to add the design.

We can add a text input to enter the key and decrypt the page

A couple of questions:

• The decryption takes ~10 minutes, will we be able to show a progress bar? Or should we think show a loading indicator that doesn't show progress?
• Can the user leave the page after clicking the button? What would happen if they do?
• If they close the page and come back to it can we show the state (still decrypting…)?
• Should we enforce that the key cannot be entered till the end of voting?

Considering this task as a best-case scenario - assuming the results come back fast enough. We'll consider the progress bar if things don't pan out as expected.

Thanks for the designs @Prtksxna

• The decryption takes ~10 minutes, will we be able to show a progress bar? Or should we think show a loading indicator that doesn't show progress?

We're hoping this won't be the case after T271824: Improve the efficiency of vote decryption during tallying - would it be OK for us to do this as a separate task if it is still a problem?

• Can the user leave the page after clicking the button? What would happen if they do?
• If they close the page and come back to it can we show the state (still decrypting…)?

Similar to the above, we're hoping that it will be like a normal form submission with results loading quickly. So it would work like any form - if you leave after submitting, you'll have to go back and resubmit. As above, could we move this to a future task if still a problem?

• Should we enforce that the key cannot be entered till the end of voting?

I agree we should. The TallyPage already has handling for this: https://gerrit.wikimedia.org/r/plugins/gitiles/mediawiki/extensions/SecurePoll/+/ddc4e600cfeefd847c1d498eab7df289364e7bf9/includes/Pages/TallyPage.php#46

In T271830#6748774, @Tchanders wrote:

We're hoping this won't be the case after T271824: Improve the efficiency of vote decryption during tallying - would it be OK for us to do this as a separate task if it is still a problem?
…
Similar to the above, we're hoping that it will be like a normal form submission with results loading quickly. So it would work like any form - if you leave after submitting, you'll have to go back and resubmit. As above, could we move this to a future task if still a problem?

Yep, that makes sense.

Change 658440 had a related patch set uploaded (by Tchanders; owner: Tchanders):
[mediawiki/extensions/SecurePoll@master] Allow decrption key to be entered via tally page

https://gerrit.wikimedia.org/r/658440

Here's how the form looks in different situations:

Unencrypted election (unchanged by this task):

Encrypted election with a key in the database (unchanged by this task):

Encrypted election with no key in the database:

Before this task	After this task

Now that the forms are more complicated, they should probably be one form rather than repeating the fields. Have filed T272916 for this.

Change 658440 merged by jenkins-bot:
[mediawiki/extensions/SecurePoll@master] Allow decrption key to be entered via tally page

https://gerrit.wikimedia.org/r/658440

You can enter the private key on the tally page and have it decrypt the results either directly from the DB or from an XML dump you already have.

When on the tally page for election x, if you upload an XML dump it does not validate that the dump is from election x. If the dump is actually from election y (and you have the private key for election y) you can use the form to decrypt election y. You could perhaps use this as a workaround to decrypt election y if you are no longer an election admin for y. But, you would still need to have access to the private key for election y, so you could decrypt the XML dump using other tools if you were so inclined.

I also tested tallying unencrypted results and encrypted results where the private key already exists on the server.

Test Environment: local docker SecurePoll 2.0.0 (3534347) 07:48, 1 February 2021.