[2.0.28] Production statistics showing wrong item counts

[sben]

Issue:

The statistics are showing wrong values for the production statistics, smaller values than the real ones.
My save has about 175 hours.
The statistic shows smaller values for some items in "Produced in last 250 hours" than in "Produced in last 1000 hours" - this is obviously wrong.


Relevant section of my journey on factorio versions:

- ...older versions...
- 2.0.23
- 2.0.26
- 2.0.28

More in detail

The mismatch is present at several different items,
I am taking for example the number of Epic quality module 2 produced on Vulcanus.

Last save from 2.0.23 shows:
- 4 produced in the last 10 hours
- 4 produced in the last 50 hours
- 4 produced in the last 250 hours
- 3 produced in the last 1000 hours !! should be 4
- 4 produced All time

<here I switched to 2.0.26, then 2.0.28>

Current save from 2.0.28 shows:
- 5 produced in the last 10 hours (this is ok since first module was produced 13 hours ago)
- 8 produced in the last 50 hours
- 7 produced in the last 250 hours !! should be 8
- 6 produced in the last 1000 hours !! should be 8
- 8 produced All time


So this must be happened either in 2.0.26 or 2.0.28,

There was a related bugfix 121799 released in 2.0.25 for accumulated science throughput, but that did not apply here.

[Loewchen]

Post the save files please, see 3638.

[sben]

The save is available at: https://drive.google.com/file/d/19esCUi ... sp=sharing

[Hares]

See also:

• 122410: [2.0.15] Production tab not reporting correct number of created legendary tier 3 quality modules
• 120132: [Genhis][2.0.15] Weird production stats for recycling recipes -- Marked as resolved

[boskid]

Duplicate 125025

[sben]

From the same save file:

An example that shows much more impact:
Uncommon iron plate produced on vulcanus

- 146 plates in last 1 hour
- 98 plates in last 10 hours
- 36 plates in last 50 hours
- no plates in last 250 hours
- no plates in last 1000 hours
- 146 plates All time

[boskid]

- 146 plates in last 1 hour
- 98 plates in last 10 hours
- 36 plates in last 50 hours
- no plates in last 250 hours
- no plates in last 1000 hours
- 146 plates All time

That still matches the issue that was fixed by 125025 and as such remains a duplicate: statistics are collected in samples and there is one special sample: one that is being created and will get commited to the statistics at specific tick. In 120351 i removed that sample from being shown in the statistics as it was affecting electric network statistics in a clearly visible way but that change affects all other statistics as well. In those longer time frames, the items produced are still held by the sample accumulator which is why they are not visible yet: they will become visible once the sample is commited. In 2.0.29 this accumulator will be always included as the proper fix for the original issue was to adjust weight of one of the samples so the effective measurement period is correct.