Help talk:CS1 errors/Archive 2

Archive 1

Archive 2

Archive 3

Archive 4

Test and implementation processes

As someone who is familiar with professional implementation of computer code, I am familiar with finding bugs in multi-levelled code (down to kernal level), but for someone who is not they would have be pulling their hair out yesterday, and would have had no idea how to work out if it was a problem they had caused or a change elsewhere. While I appreciate that Wikipedia is an encyclopaedia that anyone can edit what are the test and implementation processes that for changing CS1 code?

I am deeply concerned that changes are being made to the functionality of the {{citation}} interfaces for which there have been no RfCs to agree the change in the functionality. This should be the first part change control process so that a functional specification can be generated .... But I assume I am teaching my mother how to suck eggs, because I assume that anyone messing around with this code is familiar with how user specification are drawn up; How a functional specification is generated from that user specification; how code is written from that functional specification and how tests are written to test against the functional specification to test the changes before implantation takes place. Luckily thanks to the history mechanism there is little need for a formal roll back procedure, but there should be a widely advertised forum where implementation errors can be discussed before a roll back takes place (as of course one has to identify cases of false positives). -- PBS (talk) 12:29, 25 August 2014 (UTC)

I have asked before for some sort of specification or style guide for CS1. The community have thus far declined. Because you are familiar with software implementation, will you write a specification for us?

In lieu of a formal process, whenever I make changes to Module:Citation/CS1, or to templates outside of this project, I always talk about it. Always. For CS1, changes are always made to the sandbox first; postings where I discuss the changes are usually here (unless started by someone else in some other place). Before I update the live module, I always post a notification here and at WT:AWB with links to the relevant discussions at least a week in advance of the update. All of this so that editors can check my work. This simple set of procedures is mostly effective. It doesn't always work as yesterday demonstrates. Do you have a better process that I should follow?

—Trappist the monk (talk) 13:45, 25 August 2014 (UTC)

Cite template throwing invalid error message for date=yyyy-mm

The cite template has recently started to throw an error message for date=yyyy-mm. This is incorrect and needs to be fixed as the outcome of the RfC on this (Wikipedia talk:Manual of Style/Dates and numbers/Archive 146#Rfc: Is YYYY-MM an acceptable date format.3F_Part_4) was not to disallow this format. Also, it explicitly states that mass changes should be avoided, and if the cite template doesn't get fixed mass changes is exactly what will happen now. --Matthiaspaul (talk) 12:51, 14 October 2014 (UTC)

I think that the strike-through will just confuse people. What does it actually mean? Right here, right now, CS1 considers xxxx-xx dates to be erroneous. An editor sees an error message and clicks the Help link. The condition in the article is described by text that has been struck through. How is the editor supposed to interpret that? The purpose of Help:CS1 errors is to completely and accurately describe the error messages' meanings. To do otherwise is a disservice to the reader.

—Trappist the monk (talk) 13:16, 14 October 2014 (UTC)

Well, running into this I as a reader would interpret it as something in transition and simply not act on it. If I was curious, I would look at the talk page or in the history. Of course, the obvious fix to the problem is to just fix the module not to throw this error any more. If you think it can't be fixed easily, we could also add a note explaining the situation. --Matthiaspaul (talk) 14:06, 14 October 2014 (UTC)

I'm sure that's true for an editor of your experience level. For others, perhaps not. Those who come to this help page are generally seeking answers, not more questions. It doesn't matter that there is dispute about the presence of an error message; it does matter that an error message exists and editors who come here want to know what it means and what to do about it. The strike through should be removed.

—Trappist the monk (talk) 15:44, 14 October 2014 (UTC)

How is this for a compromise? or this? —PC-XT+ 17:54, 14 October 2014 (UTC) 18:16, 14 October 2014 (UTC)

Isn't 'warn' a bit strong? The closing admin "[recommended] ... [that] mass changes [shouldn't be] made purely on the basis of [the] RfC." (emphasis mine)

—Trappist the monk (talk) 21:44, 14 October 2014 (UTC)

 Done Yes, it should be close to the wording of the RfC result. I made the change. —PC-XT+ 23:00, 14 October 2014 (UTC)