Eds to TCP
A little while back (16:29, 21 February 2015) you made an edit to the lead of transmission control protocol that removed a statement that covered the fact that TCP's idea of reliability is not quite what most would expect, as in it is not reliable delivery, but is actually deliver or notify. I'm not sure why I failed to notice the change closer to the time, but I have now. Anyway, rather than just revert on that specific issue, i.e. chance "TCP provides reliable, ordered, and error-checked delivery of a stream of octets..." back to "TCP provides reliable, ordered and error-checked delivery (or notification of failure to deliver) of a stream of octets..." I thought I'd try to discuss it with you first.
I know that part of the problem is that I work on a very specific, niche application of PSNs in realtime systems. And the nicheness of that application results in me having a somewhat distorted perspective, wrt everybody else’s. But in that specific situation there is a very significant distinction between “reliable delivery” and “deliver or notify”, in that notification of failure to deliver has almost no value. I have got something on the issue in at Reliability (computer networking)#Reliable delivery in real-time systems, which I hope explains the issue properly.
Now I wouldn't propose to discuss this issues in any detail in the TCP article. I think that, unless you come at the problem from the real-time perspective, TCP is, statistically speaking, reliable. But from the real-time perspective “there are lies, dammed lies, and statistics”, as Sam Clemens credited to Disraeli. Hence, I do think it's worth at least that much coverage, i.e. "(or notification of failure to deliver)".
So if you fundamentally disagree on this point, as it appears from the deletion of the note, could you give some rationale for that?
- I think you are referring to the lede of the article. In rereading I still stand by that deletion. The lede of a article should summarize the article, not introduce fine nuances that cannot be understood in a handful of words in a parenthesis. If you deem this nuance even important in the scope of the article, then it should be discussed in a section of the article body. Many words in this lede or other technical articles cannot be explained or qualified sufficiently in the lead and the reader should not be confused with short phrases that probably need even further explanation. Kbrose (talk) 16:44, 2 March 2015 (UTC)
Your ping edits
- whether or not they are "portable", whatever that means, is totally irrelevant. The example illustrates what the utility does, and is not a manual how to use it. What other options other operating systems have is also not interesting to WP. WP is not a how-to manual. Your comments about wrong usage is also inappropriate, as the example comes from a Unix-like system and that is stated, which kind really is irrelevant. Kbrose (talk) 23:01, 9 March 2015 (UTC)
- Please try to be less confusing. If you insist in hiding the non-portable details, we would need to completely remove the example. Is this what you intend? Schily (talk) 23:07, 9 March 2015 (UTC)
I'd wondered whether to ask you to check International System of Units so I was glad to see you work on it anyway and glad again to see you deal with some points that had bothered me. Thanks. NebY (talk) 11:14, 16 March 2015 (UTC)