Charts/data with each category error counts over time

[stefanb]

It could give an interesting visualisation, indicating massive vandalism etc.
Maybe also plotting the time to fix an error (min, max, mean, stddev...)

Even just providing the raw, non agregate data would be great, so visualisation can be done elsewhere.
ErrorID,ErrorCategoryID,RelationID,details,detectedTimestamp,fixedTimestamp
maybe even with links to OSM changeset IDs if available via #8 (both for breaking and fixing changeset).

Maybe add even some geographical context (bbox, or country), so that users can filter the data and adjust their pager duty alerting accordingly :)

It could be done by scraping the data as indicated in #13, but some isues may be lost if they are detected and fixed in short period between scrapes.

[DavidKarlas]

I really want to fix #8 for many reasons, still didn't spend enough time thinking about it to figure out how doable it is...

This issue doesn't sound too hard, I just need to find some scalable storage where I dump changes as I detect them...

Only problem that I see is that I will need to make whole service more reliable, today it happens that it goes offline for hours before I notice... power goes out, or updates... This feature puts some pressure on being online 24/7... :)

Whole thing is designed in way that multiple analyzers can run in parallel for high availability and all upload to single issues.json so even if 1 goes offline work moves on, and website is just issues.json beautifier...