Fri, Feb 14
How will the relative bits work with translation and localization if this is truly configurable per wiki?
Thu, Feb 13
This is a good question. My first instinct would be to do single inserts inside a database transaction.
What would the variable look like? Does it include the UI language and the actual "offset" value like number of days from now to expire? Is this an array then?
I think I like that idea. Any downsides we can foresee?
Tue, Feb 11
Their response: "Are you using an up to date version of calibre?" Eh....
If we uncover what we think might be bugs or inconsistencies with Calibre's renderings, it would be cool to create issues or bug reports for them. We probably can't fix it but we can help with a bit of research.
Mon, Feb 10
I salute you!
Awesome. I hope this shows benefits for the reliability. Initial signs look good.
Fri, Feb 7
@Marostegui Thanks for your help with this!
Thu, Feb 6
Thanks for clarifying that, y'all.
@Marostegui Yes, it can be replicated as long as the watchlist table is replicated. This table only has value with the watchlist table data.
Is there anything else we need to discuss before getting this change on the calendar?
Wed, Feb 5
Thanks for the feedback y'all. I like the approach that @Anomie detailed above. It makes sense and seems much safer.
I don't think we want to get into the work to add another column to the watchlist table.
Tue, Feb 4
Thanks! Maybe we should make a separate task for fixing the mobi export?
Mon, Feb 3
Agreed. The downtime emails are pretty much back to their original state.
Sat, Feb 1
Thu, Jan 23
Wed, Jan 22
Jan 16 2020
TIL: You can't merge pastes with tasks. But I made a nice comment on the paste: https://phabricator.wikimedia.org/P10183#59317
Should we merge this with: P10183
This has a patch now: https://github.com/wikimedia/WhoWroteThat/pull/145
Jan 15 2020
@ifried I asked Leon to go ahead and start on this because it was becoming really problematic and nearly unusable.
Jan 14 2020
I applaud that effort personally. However, the users of this tool rely on that data. They are quite literally trying to fingerprint a user with this information. It's clear from that post that they want to thwart fingerprinting (which, again, I applaud).
Jan 8 2020
I will keep an eye on this and create a new task if more problems arise.
Dec 16 2019
@DannyS712 This is a very old task. We've deprioritized any work on Multiblocks by our team for the foreseeable future. It was originally an idea about extending Partial Blocks.
Dec 13 2019
Thalia is correct here. Maybe I didn't capture the details correctly in the initial tasks I wrote. Still, we did have the conversation and this is the result. I'll work to ensure we document things better in the future. I might get annoying about it. :)
Dec 10 2019
I had a few random thoughts about this:
- The DELETEs need to ensure the relevant rows from both tables are deleted.
- We might need some understanding of the transaction scenario here. I imagine we'll be doing the delete in a single query. That means that if the query fails, none of the appropriate expiring items will be deleted. How should we recover? Try again? Loop through each one and DELETE individually? One of these might be better or worse for performance.
- This table is guaranteed to be smaller than the existing watchlist table.
- One smaller wikis, I suppose there's the possibility that this table might be completely empty from time to time. That chance exists on a larger wiki but the probability approaches zero pretty quickly.
Dec 9 2019
I think there are also performance gains to be had from a smaller overall table when doing large bulk deletes. My thinking on that might be outdated.
Dec 6 2019
As Harumi's manager, I approve.
Nov 23 2019
As Max's former manager, I endorse his access to these resources under and NDA.
Nov 22 2019
My opinion is that non-org is fine. Thanks for tracking down all the details on this.
Nov 21 2019
Nov 15 2019
I had a very similar reaction to the data. I made this point when we first talking about this task. I think this is an initial pass at the data and probably gives us a sense of the upper bounds of what's possible but isn't actually realistic.
Nov 14 2019
Nov 13 2019
That works for me.
@dbarratt I see. That makes sense to not try to overload the API that way.
Nov 4 2019
Nov 1 2019
That works for me. It's about degrees of risk here. If we feel that prioritizing it as you've described is acceptable level of risk, then I think that's fine. There is no risk-free option except to completely remove the feature so managing the risk with clear eyes is the best approach to me.
@dom_walden is spot-on I think. The behavior he describes is definitely what we talked about but maybe it got mixed up somewhere.
Oct 31 2019
One way to output this data would be as a histogram with the average time per month and per wiki. It might be helpful if we see big shifts in the average time, maybe a policy or procedure changed on a given wiki. Or, even more interesting, maybe the tool changed somehow.
Oct 29 2019
@MaxSem Thanks for that. Turns out that it's not specifically just using EL but having whitelisted schemas whose data is kept for longer than the normal retention period. We do not have any of these special schemas.
Oct 28 2019
Thanks y'all for working to get this one right.
Oct 25 2019
@mforns Thanks for that info. It was very helpful.
Oct 24 2019
Is there a way for me to discover the schemas that this team has created? Our institutional memory is fuzzy. It'd be great if I had an authoritative source of the schemas we created which are still gathering data.
Oct 23 2019
@DannyS712 You've claimed this task so we've not been working on it further. Do you intend to come back to it? If so, we will wait but if you've moved on, we can jump in and finish up the work.
You will be reimbursed the $5 fee if you make an expense report.
Oct 22 2019
I can +2 the base patch.
Oct 21 2019
Oct 17 2019
This is very cool. Thanks for sharing @dom_walden!
@Samwilson Thanks for clarifying that. I'm glad you put the decay in there.
Oct 16 2019
@dom_walden Thanks for that detail. As for backoff, I was thinking that in other retry scenarios I've created, the delay between tries gets longer and longer the more times you try. This is in an effort to help the downstream system recover if it's under strain. I'm assuming we aren't doing that.
Four retries seems fine from my perspective. What's the interval? Does it degrade or backoff?
I wanted to share a general bit of guidance with regards to these "security" fixes. I think it would be helpful for us to think more in terms of defensive coding instead of secure or insecure or even more specifically whether we trust the source of the data or not.
Oct 11 2019
I spoke to a friend who still works in this area and they said that spam detection and management is in freefall at Yahoo/AOL right now. They are rapidly defunding that part of the business and many automated operations are happening with little human oversight.
Oct 9 2019
@ifried That's fine. I wanted to scope the responses you're expecting.
Do any of these considerations change or have unique limitations in the mobile context?
The PR is merged but it appears to me like this might be one of the things that the Mozilla folks were concerned about. The documentation even says, "Accepts raw HTML."
Oct 8 2019
Oct 7 2019
I guess I was thinking about a template library and not string literals. I misspoke.
Oof. I think we knew this was a risk, right? Looks like we need to do sanitization or go toward some ES6 template sort of thing like @MusikAnimal was proposing at some point.
I've +2ed this patch.
Pages with the Most Revisions is the report that fails on occasion. We will leave it in because it does work sometimes and some people get value.
Oct 3 2019
Thanks, y'all. Having that priority will help us know when to raise potential effort/sizing issues about the could haves.
I don't know of anything concrete about what might change with the Desktop Refresh. I think we should proceed with the current state and just understand that it may need to adjust in the future.
This seems like a nice feature.
Oct 1 2019
Who is the primary audience for these reports? Where do they end up?
Sep 30 2019
Moving back to Ready as we are changing the design in general.
Sep 19 2019
The limit is 10 pages.
Thanks @nettrom_WMF. We will take a look. I don't know if we have this off the top of my head.
Sep 17 2019
Works correctly for me as well. Chrome 77 on Mac.
Sep 12 2019
Thanks @Samwilson! That's very helpful.