Bug 321842

Summary:	false error indicated in incomplete row/column
Product:	[Applications] picmi	Reporter:	dmatt <xdarkmatter>
Component:	general	Assignee:	Jakob Gruber <jakob>
Status:	RESOLVED FIXED
Severity:	normal
Priority:	NOR
Version:	2.0.0
Target Milestone:	---
Platform:	Ubuntu
OS:	Linux
Latest Commit:	0c1603117826f37d8520cbf9951e63379f58f5ee	Version Fixed In:	4.12
Attachments:	example of column evaluated as all wrong bug manifestation example

Description dmatt 2013-07-01 22:22:46 UTC

logic checking individual streaks marks whole row/column as incorrectly filled in, even though only individual streak has was not completely checked

Reproducible: Always

Steps to Reproduce:
1.completely correctly fill in any row/column with at least streak longer than 2
2. uncheck any cell inside of long streak (split it into two segments)
3.
Actual Results:  
all numbers for given row/column turn bold indicating it was filled incorrectly

Expected Results:  
only streak/segment which has unchecked cell should be bold

Comment 1 Jakob Gruber 2013-07-02 07:57:59 UTC

In some cases, marking all numbers invalid if one cell is unmarked is intended to avoid leaking information or giving a false impression of actually having completed a line. To be honest, it's been a while since I wrote this logic and I don't remember exactly though.
It would be helpful if you could post or attach some specific examples.

Comment 2 dmatt 2013-07-02 21:38:44 UTC

Created attachment 80904 [details]
example of  column evaluated as all wrong

Comment 3 dmatt 2013-07-02 22:15:23 UTC

Created attachment 80905 [details]
bug manifestation example

Picture (1): 2 cells are not marked, indication correct
Picture (2): (one more cell was correctly marked), indication incorrect (all streaks show as wrong)
Picture (1): (remaining cell was correctly marked), indication again correct

If just one streak is incomplete and empty cell(s) are not on the edge of streak, , indication is wrong.

Not very common during gameplay, but pretty confusing when it happens

Comment 4 Jakob Gruber 2013-07-03 08:01:23 UTC

Right, thanks. I've been thinking about adding a couple of unit tests for the streak algorithm, so I'll probably start off with that and look at this particular problem afterwards.
As always, patches welcome!

Comment 5 Jakob Gruber 2013-08-13 21:42:42 UTC

Hi there,

the patch series culminating in commit 0c16031178 should fix this particular issue.
Please let me know if any further issues pop up.