#dev | Logs for 2017-06-03

« return
[02:19:57] <TheMightyBuzzard> there's not really a preferred guideline. if it works we tend not to complain.
[03:49:50] -!- SoyGuest64809 has quit [Ping timeout: 276 seconds]
[03:50:38] -!- SoyGuest64809 [SoyGuest64809!~DECbot@90-244-262-780.prtg.in.frontiernet.net] has joined #dev
[09:58:42] <SoyGuest77599> devs, please review: https://github.com
[09:58:54] <SoyGuest77599> freshly rebased just now.
[09:59:20] <SoyGuest77599> THEre's tons more, but I want to make sure I'm moving things in the right direction.
[09:59:46] <SoyGuest77599> Personally, I prefer using ';' as a parameter separator, because of '&' being overloaded.
[10:01:14] SoyGuest77599 is now known as FatPhil
[10:01:17] -!- FatPhil has quit [Changing host]
[10:01:17] -!- FatPhil [FatPhil!phil@Soylent/Staff/Editor/FatPhil] has joined #dev
[10:30:02] <TheMightyBuzzard> FatPhil, re: your PR. Why? & is for urls, &amp; is for displaying. Even on the chance it doesn't break every place you changed it, we gain nothing.
[10:53:35] <FatPhil> I think you need more coffee
[10:54:30] <TheMightyBuzzard> well that's a given
[10:55:31] <FatPhil> In href= and src=, the & character must be entity escaped, otherwise it looks like it's beginning an entity itself
[10:55:51] <FatPhil> http://validator.w3.org
[10:56:03] <FatPhil> Line 661, Column 48: cannot generate system identifier for general entity "start"
[10:56:13] <FatPhil> <b><a href="//soylentnews.org/search.pl?sort=2&start=15">Older Articles</a></b>
[10:56:32] <FatPhil> An entity reference was found in the document, but there is no reference by that name defined. Often this is caused by misspelling the reference name, unencoded ampersands, or by leaving off the trailing semicolon (;). The most common cause of this error is unencoded ampersands in URLs as described by the WDG in "Ampersands in URLs".
[10:56:52] <FatPhil> which -> http://www.htmlhelp.com
[10:58:08] <FatPhil> line numbers may of course vary, as we may have slightly different views of the page.
[11:00:06] <TheMightyBuzzard> unless there's an RFC saying you should always do it this way, i disagree with the validator. entities in urls is insanity. browsers should not parse them.
[11:00:09] <FatPhil> run the validator over a story page, and the error count jumps into the hundreds (depending on the number of comments on the page)
[11:00:39] <FatPhil> w3 wrote that validator, I think they know their own standard
[11:01:12] * TheMightyBuzzard shurgs
[11:01:46] <TheMightyBuzzard> doesn't mean much. unless they're pointing to an RFC, it's just their opinion.
[11:02:38] <FatPhil> URLs in HTML do have their entities interpreted, do you really want me to have to hunt out the part of the DTD that specifies that.
[11:04:08] <FatPhil> href %URI; #IMPLIED -- URI for linked resource --
[11:04:14] <TheMightyBuzzard> i believe it's possible but i'm not gonna potentially break a ton of our code to make them happy unless it's part of an official spec
[11:04:24] <FatPhil> where %URI is
[11:04:25] <FatPhil> <!ENTITY % URI "CDATA"
[11:04:26] <FatPhil> -- a Uniform Resource Identifier,
[11:04:26] <FatPhil> see [URI]
[11:04:26] <FatPhil> -->
[11:04:45] <FatPhil> where CDATA is
[11:04:46] <FatPhil> CDATA is a sequence of characters from the document character set and may include character entities. User agents should interpret attribute values as follows:
[11:04:49] <FatPhil> Replace character entities with characters,
[11:04:52] <FatPhil> ...
[11:05:05] <FatPhil> **** Replace character entities with characters, **** <--- THIS BIT HERE
[11:05:29] <TheMightyBuzzard> blerg, that's gonna break a lot of shit
[11:06:00] <FatPhil> Show me one specific thing that it will break.
[11:06:48] <TheMightyBuzzard> i'll need to bytram it first. but we do a shitpot of conversions of entities and exclusion of entities in the code.
[11:07:15] <FatPhil> this is only on the output side
[11:07:38] <FatPhil> we are outputting invalid HTML, and that's a now-proven fact.
[11:07:45] * TheMightyBuzzard ponders
[11:11:07] <TheMightyBuzzard> aight, seems massively insane to me to have browsers parse entities but if the whole world wants to be retarded who am i to say we should be intelligent
[11:11:53] <FatPhil> It seems massively insane to you for browsers to follow the standard?
[11:12:09] <TheMightyBuzzard> it seems massively insane to put that in the standard to begin with
[11:12:09] <FatPhil> Did you work for MS on the IE team?
[11:12:35] <TheMightyBuzzard> why in the jumped up monkey would anyone want to open up that particular can of worms?
[11:12:49] <FatPhil> How would you pass a copyright symbol to a GET/PUT request over an ASCII transport?
[11:13:23] <TheMightyBuzzard> are there any ascii-only browsers left?
[11:13:30] <TheMightyBuzzard> or servers?
[11:14:28] <FatPhil> When the standard was written, there were.
[11:15:12] <TheMightyBuzzard> ahhh, so it's legacy cruft that needs to diaf then
[11:15:29] <FatPhil> No, it's called the standard that we're supposed to be conforming to.
[11:15:45] <TheMightyBuzzard> the two are in no way mutually exclusive
[11:16:09] <FatPhil> If you think it's legacy, then don't use HTML 4,01.
[11:16:13] <FatPhil> Or XHTML
[11:16:17] <FatPhil> Or HTML 5
[11:16:26] <FatPhil> Because the same clause is in all of those standards.
[11:16:44] <TheMightyBuzzard> it's very much legacy unless there are still browsers or servers that do not offer unicode support
[11:18:07] <FatPhil> You did work for MS on the IE team - you're basically saying "standards be damned"
[11:18:33] <TheMightyBuzzard> i'm saying standards should not be standards just because they're standards. they should require logic.
[11:18:52] <TheMightyBuzzard> and when the reason for something disappears, so should it.
[11:19:00] <FatPhil> Entities avoid lookalike spoofing, there's still plenty of value in them
[11:19:29] <TheMightyBuzzard> and they cause parser-writing headaches and exploits as well.
[11:19:38] <FatPhil> because, as I said, your argument boils down to "fuck entities, I can just use the unicode character"
[11:20:14] <FatPhil> They certainly cause parser-writing headaches for people who chose to not *follow the standard*.
[11:20:31] <TheMightyBuzzard> no... they cause the exact opposite.
[11:20:32] <FatPhil> If you consider that a problem, then you must consider yourself part of the problem.
[11:21:18] <TheMightyBuzzard> dude, this is not a personal argument and you seem to be taking that way. we'll pick it up another time when you've calmed down.
[11:21:31] <FatPhil> It's an engineering argument.
[11:22:02] <FatPhil> I code to specs. I write parsers.
[11:38:33] <Bytram> My quick, under-caffeinated take on Named character entities in URIs: based on my experience, (1) I can think of no case where I've seen a browser reject a URI which contained a named character entity in the URL. (2) I have seen a shit-load of errors from all manner of validators that I have used which have complained about 'naked' ampersands in URIs. (3) Given that ampersands ARE permitted in URIs, browsers will attempt to resolve a NCE
[11:38:34] <Bytram> following an ampersand which (a) can be circumvented by using '&amp;' to *produce* an ampersand, and (b) provides a small, but not impossible, attack surface where a naked ampersand can be followed by text which ends up be manipulated into an NCE.
[11:39:10] <Bytram> BTW, the REAL reason I stopped in here is to report an error that I predicted would occur...
[11:40:30] <Bytram> When subscriptions are processed, and the only subscription(s) for a day are "grants", i.e. have no $ amount, invariably it is the case that slashd will generate an error, to wit:
[11:40:31] <Bytram> SoylentNews Slashd Error Alert!
[11:40:31] <Bytram> SLASHD line 165 produced the following error:
[11:40:31] <Bytram> subscribemail.pl odd exit (status 25, signal 0)
[11:40:47] <Bytram> ----------------------------------------------------------------------
[11:40:47] <Bytram> You have received this message because you subscribed to it
[11:40:47] <Bytram> on SoylentNews. To stop receiving this and other
[11:40:47] <Bytram> messages from SoylentNews, or to add more messages
[11:40:47] <Bytram> or change your preferences, please go to your user page.
[11:40:47] <Bytram> https://soylentnews.org/my/messages
[11:40:49] <Bytram> You can log in and change your preferences from there.
[11:42:17] <Bytram> I've reported this in the past, but hopefully with some new eyes looking at the code, this can be addressed. I'd looked at the code many many moons ago, but it was just past my kenn to understand why it was going south.
[11:42:59] <Bytram> sadly, no time to chat today... need to get going for work and to the store before work to get stuff for a graduation tomorrow that I was invited to yesterday. :/