Template stubs
editI disagree with you chaging stubs as you are. --Spinboy 17:15, 2 Mar 2005 (UTC)
- On what grounds? It is not advisable to change the metastub or metapicstub templates sinice they are linked to too many articles, which may cause a DoS-style attack. Because of this, there is no point in keeping the meta templates, because even if we wanted to change the look of all stubs, we couldn't (because of the risk of DoS-style attack). So why not remove it if it will increase performance? --jag123 19:35, 2 Mar 2005 (UTC)
- On the grounds that no consensus has been reached. They may want us to be bold, but if it's controversial, you should wait for consensus first. --Spinboy 19:42, 2 Mar 2005 (UTC)
- A database developer gave his input regarding the matter. Do you want consensus from the other developers that it's okay, or from the community that supports whether or not that developer knows what he's talking about? People can disagree on whether the templates should be changed, but that doesn't mean what the developer said is wrong. By the way, it's really silly to revert my changes because "there's no consensus", because there isn't any consensus that the metastub or metapicstub should be kept either. --jag123 19:50, 2 Mar 2005 (UTC)
- That's right, so if there's no consensus, then you should leave it as-is. --Spinboy 19:58, 2 Mar 2005 (UTC)
- Now that's it been changed, you think it's okay to revert my changes, citing them as vandalism, even though there is no consensus for keeping the meta templates? How does this affect you anyway? --jag123 20:03, 2 Mar 2005 (UTC)
- How does it affect you? You think it's okay to go and make the changes, citing some stupid page you put up, even though there is no consensus for getting rid of the meta templates? How does this affect you anyway? --Spinboy 20:19, 2 Mar 2005 (UTC)
- It affects me because every page using a template that uses the meta-template makes more calls than necessary to the database, which hurts Wikipedia performance. Perhaps you like having a slow Wikipedia, but I don't. If this helps in some small way, then great. I put up this "stupid" page because I didn't think putting all those links in the edit summary was appropriate. I don't plan to get rid of the meta templates, I'm just removing them from all of the stub templates, which affects absolutely no one in a negative way, except someone who wants to change the look of all the stub templates easily. Someone who wants to change the look of all stubs can go through all templates, like I just did. People can keep using the meta-templates so a consistent look and feel is maintained; but they can be subst: later (which is what I did).
- All you've done is dodge my questions and cite consensus, as if the lack of it means you're right. You're really not helping anything. --jag123 20:39, 2 Mar 2005 (UTC)
- That doesn't affect you at all. It doesn't affect me. It just bothers me that you assume your way is the only way. --Spinboy 06:36, 3 Mar 2005 (UTC)
I disagree with the tone Spinboy has been taking, but I agree with his position. This is the sort of thing that's liable to be a pain to change back. Discussion is still active; no consensus has been reached. The WP:BB page you're so fond of referencing contains a section asking one to not be quite so bold with edits that would affect a lot of people. This is one of them. "Be bold" doesn't mean "be bold in declaring your viewpoint the victor of a debate and start making changes". grendel|khan 02:35, 2005 Mar 4 (UTC)
- I never declared my viewpoint to be the victor, but unless there is a real dispute (not just people thinking that it can't be that big of a deal) about whether using meta-templates causes performance issues (and coming from a developer, I trust it to be true), then it certainly can't hurt that the templates stay this way until a consensus is reached. I still don't see how this affects an editor involved in stubbing, much less the average editor, and Spinboy certainly hasn't helped on clarifiying why he cares. If the community decides to keep the use of meta-templates, then I will revert them all back myself. I ran this (the conversion) by a developer before beginning, and I wasn't discouraged or told not to bother. Quite the opposite, actually. Until a consensus is reached, anyone is entitled to do what they wish. I am no more right or wrong in converting them than Spinboy was in reverting them. I have absolutely no emotional attachment to these templates, and honestly, really couldn't care less if they change or don't. They were described as harmful, and I changed them to help and that's all there is to it, so let's take it easy with the warnings/statements/insinuations that I'm acting in bad faith here or that I think I'm right/above/more entitled/etc than any other editor. --jag123 05:05, 4 Mar 2005 (UTC)
Conversions
editYou wrote:
- Note: this assertion has also been contested. No consensus has yet formed on the usefulness or harmfulness of metastubs.
No consensus formed, yet you're acting as though one was formed. I guess WP:BB, but still. Cburnett 18:54, 2 Mar 2005 (UTC)
- I added that note. I'm not acting like a concensus was formed either way. (I haven't changed any templates to use or not user meta-templates, and I don't have a strong opinion either way.) I think you have me confused with some of these other folks. Regardless, I think there needs to be a Wikipedia-wide survey to determine what the consensus actually is. – Quadell (talk) (sleuth) 21:48, Mar 2, 2005 (UTC)
Consensus from who? Wikipedia's main database developer gave his input on this matter. The community needs to reach a consensus on whether or not the developer knows what he's talking about? PS: I didn't add the note on consensus either, and yes, I am being bold. --jag123 19:35, 2 Mar 2005 (UTC)
Technical Good/bad
editHere's a quick summary of template technical bits:
- Every template use adds an extra database query for each page build.
- Templates within templates double that extra load, compared to a single template.
- Each change to a template causes purging of all pages using it form the caches and cuases some lag, and possibly out of date versions of articles, watchlists and recent changes. The cost of this varies depending on the number of pages involved. Relatively few pages at a time are mostly invisible, large numbers become very visible because they exceed the wait time which is used to mask smaller purge times.
- While the general load can be solved by buying more servers, the lag part can't be, because that affects every server at the same time, as all have to do the work.
- Templates can save a lot of human work.
- Substituting the template content at the time the page is saved instead of including it dynamically when the page is viewed avoids all of the negative load and speed effects.
This suggests the following general thoughts on when templates are most and least appropriate as an approach to a particular editing task:
- If the content is changed very frequently, it's quite likely that a lot of human work is being saved, perhaps enough to make it worth doing with a template if it's important that everyone see the latest version. The main page templates which change daily are an excellent example of this type of template - regular changes and few pages being updated each time they are changed.
- If the content changes infrequently or it doesn't matter much if it's a few versions old, there's less reason to take the extra load cost. Things like the stub notice are a good example - a stub stays a stub regardless of the exact text of the stub notice at any given time.
- If a template or meta template affects relatively few pages, it's relatively unlikely to cause the lag and out of date versions issue, so it's less harmful than a template used on a vast number of pages. A template for a stub notice, not broken down by letter of the alphabet, would be a good example of one which causes a lot of visible delay and out of date content every time it's changed, one which is a bad candidate for template use.
- Templates within templates increase the negative effects and the first level templates already save a lot of work, so in general, meta templates make the costs greater and deliver lower benefit than other templates.
There's no one answer for every template situation, but in general: dodging templates within templates and dodging templates where it's not really important if people see the latest version will lead to the best mixture of fast page loads, few visible lag times and best balance of human work saved for server load added. Jamesday 06:05, 4 Mar 2005 (UTC)