| Stevietheman | |
| Oct 30 2015, 11:41 PM |
| F2904470: excesspipes.xml | |
| Nov 1 2015, 11:58 PM |
Citation templates (FixCitationTemplates)/"Removes excess pipes" was not skipped even though it was the only fix on the page and I was skipping if "Only minor genfixes". Could you make sure this is set as a minor general fix?
If you need me to provide more info, just let me know.
@Stevietheman You can check the manual http://en.wikipedia.org/wiki/WP:AWB/GF for which fixes we consider "minor" and which we consider "MoS".
I know this. And this fix is marked as minor on that page. It says "This is a minor fix".
I have tried to distill a test case, but this issue doesn't show itself if isolated. I am convinced this happened, though, because this was the only fix made on the page, and it was supposed to be skipped.
This makes me wonder if there isn't some kind of issue with not "resetting all the flags" (or however you would like to describe it) before processing each new page. I don't have a test case to prove this, however, so this can be closed.
@Stevietheman I think if you want to achieve maximum efficiency enable ALL of "skip if minor fixes", "skip if only whitespace removed" and "skip if only casing changes". There is a chance some whitespace was remove by other function which was is marked as "minor".
I don't think that has anything to do with what I saw. The ONLY change on the page was the removal of the excess pipe.
I'm glad this wasn't closed yet, because I have gotten it to repeat, and it occurs with a basic settings file:
AWB tries to correct two instances of excess pipes in two cite templates, even though I have it set to skip minor genfixes.
https://github.com/reedy/AutoWikiBrowser/blob/master/WikiFunctions/Article/Article.cs#L1498-L1499
https://github.com/reedy/AutoWikiBrowser/blame/master/WikiFunctions/Article/Article.cs#L1498-L1499
The Parsers.FixCitationTemplates function has been marked as making "significant changes" for over 2 years (last time the call line was altered according to blame).
It seems the code does a lot more than just fixing excess pipes. So the question is... Which is actually wrong? The documentation? Or the call marking it as a significant change?
In this case, the reader sees no difference, so that makes it minor in that sense, IMHO.
I can see where the function as a whole, if it performs more visible changes (if necessitated), might be considered significant.
As the AWB user, it boils down to me making a change that feels minor, but AWB treating it as if it's significant.
Some templates are placed in tracking categories when they have unnamed undefined parameters.
Honestly, without spending a lot of time digging into what the function does, or can do...
I don't know whether it's worth making that level of granularity to the individual changes AWB makes. Changes like this are made as part of doing larger fixes, so they carry less cost at the time of doing (at least, if also making major changes). Refactoring them out to be seperate is a performance problem waiting to happen I would suspect.
You can, of course, ignore/skip these pages manually.
What's the level of false positives with this function? ie changes that come up that still feel minor?
If that is indeed, up to date, it would seem that we should be marking the fix as minor, not a major one is the correct course of action.
I can also save them. Heh. It doesn't happen a lot where this is the only change standing, and perhaps other users would see the utility of me fixing a template despite not ending up with differences for the reader. This report really is about a seeming discrepancy between the skip option I used and the change presented to me. As a former software developer, things like that naturally bother me. :)
What's the level of false positives with this function? ie changes that come up that still feel minor?
Maybe it's rare, I don't know for sure. What I do know is that whole section is marked "This is a minor fix". So maybe this can end up being called a documentation issue?
If that is indeed, up to date, it would seem that we should be marking the fix as minor, not a major one is the correct course of action.
Yes, that's pretty much where my confusion began. Either the documentation is wrong, or the software is.
Documentation is wrong, I have updated. Magioladitis made changes in February 2013 that were wrong https://en.wikipedia.org/w/index.php?title=Wikipedia:AutoWikiBrowser/General_fixes&diff=538396736&oldid=538396649
The "only minor fixes" feature is not well developed. We currently have functions/methods that may make a mixture of minor and significant changes; it may also be contextual as to whether a particular change is minor or not (e.g. some whitespace removal affect page layout, others do not). Realistically if people complain more about the feature I would remove it rather than try to determine on a line by line basis through all general fixes whether a change was never/sometimes/contextually/always significant.
We have the "only cosmetic changes" option which uses the API to see if the edit would change the HTML of the page. That's the better answer if you want to automatically skip what may be considered minor or cosmetic changes.