#ossasepia | 2019-11-01

← 2019-10-31 | 2019-11-02 → ↓

billymg: whaack: i have run into this same bug myself. the problem is html renders "escaped" html characters as their symbols. e.g. if you initially input something like "" to render a tag it will get saved as such in mysql. but when you edit the post, and the same "" is pulled out of the db and put into the text area, it will render as ""

billymg: so saving again saves the "", not the "" you originally entered

whaack: billymg: yup, that is my understanding of the problem

billymg: this might be helpful https://www.php.net/manual/en/function.htmlentities.php

billymg: running the db output through that before inserting into the <textarea> converts everything to html entities. so tags you intended as tags, e.g. , go into the <textarea> as (rendered as the desired ), and tags you intended to escape, e.g. , go into the <textarea> as &lt;p&gt; (rendered as the desired )

billymg: would certainly be a useful patch if you have the time to test further and ensure it works as expected without side effects

billymg: also provided diana_coman finds it a worthwhile use of your time i should say

whaack: billymg: ^ hah that is always a given. yes it seems like a good idea at first pass

whaack: lobbes: i am working on a solution to this http://logs.ossasepia.com/log/trilema/2019-10-31#1948993 . you don't want to put any string for the b query param that is going to match a string inside of an anchor tag. this problem occurs frequently because often the same text in a link you want to select is inside of the anchor tag: i.e. for the text 'mp writes at <a href="http://trilema.com">trilema</a>' if you try to select trilema w

whaack: ith b=trilema you will break the anchor tag instead.

ossabot: (trilema) 2019-10-31 lobbes: in other weird, I found an odd (albeit edge case) bug in the new blog selection mechanism. If you do a select that begins with "http" (e.g. "?b=http&e=#select") it will mangle the selected link in the post itself when rendered

whaack: the other weird about the new selection mechanism is that when you leave no value for e, there is no closing being added anywhere. afaik all browsers handle this by 'closing' the tag by putting a before either the next (1) the opening tag of a block element (2) the closing tag of any element that the opening tag is nested in

lobbes: whaack: aa interesting. I didn't know that about the no value for e; now that weird makes sense at least

lobbes: and nifty. I look forward to your solution (possibly this was in the #o logz already and I missed it)

whaack: lobbes: there is no solution proposed yet. the first two fixes that came to mind are to (1) make sure that you are not inside of an anchor tag when putting in the opening and (2) to make sure you put the explicitly if you hit one of the above scenarios before finding a match for the "e" value

jfw: http://logs.ossasepia.com/log/ossasepia/2019-10-31#1008349 - I had also encountered this

ossabot: Logged on 2019-10-31 18:33:26 whaack: lol til after hitting 'save draft' on a mpwp post with html characters escaped, the escaped characters are evaluated to their html characters, thus hitting 'save draft' twice in succession can alter the content of a post.

jfw: (per selection, implications are worse than noted here so far - editor page can be html-injected)

jfw: re http://logs.ossasepia.com/log/ossasepia/2019-10-31#1008357 I'd say the problem isn't html doing what it's supposed to do i.e. displaying < as a literal less-than, but wordpress failing to account for this when rendering the initial editor content, exacerbated by the browser accomodating the brokenness to some degree.

ossabot: Logged on 2019-10-31 21:05:59 billymg: whaack: i have run into this same bug myself. the problem is html renders "escaped" html characters as their symbols. e.g. if you initially input something like "" to render a tag it will get saved as such in mysql. but when you edit the post, and the same "" is pulled out of the db and put into the text area, it will render as ""

jfw: billymg, know a reason to prefer htmlentities to htmlspecialchars? fwiw I can't tell why the former even exists and the docs don't fully specify what it does

jfw: I had a couple draft patches for this and other issues; I could publish next week but will defer to diana_coman on priorities.

jfw: re selection mechanism quirks, note that the span tag insertion works at the low level of string insertion, no awareness of the HTML structure or anything, and I'd caution against trying to hack something like that onto it as it pulls in the whole mess of parsing malformed tag-soup since wordpress isn't strict to begin with.

whaack: jfw: my current strategy is to disregard the position of the match for "b" if, after the position of the match, a close tag comes before an open tag. then try 1 more time to find a match after the close tag.

whaack: (this is for fixing the problem that causes the select-display to break elements/links, i haven't done too much thinking re closing the element)

whaack: i agree with you that we cannot assume proper HTML structure. but since the sel-display is a best effort basis, i think it is the right thing to not insert a span tag when correct HTML structure would imply we'd be placing that tag inside of another tag.

whaack to sleep

spyked: http://logs.ossasepia.com/log/ossasepia/2019-10-29#1008077 <-- lulz @ "fast-food technology". I'm wondering how this translates to software in technical terms. is it just the inflation of code, or what?

ossabot: Logged on 2019-10-29 18:15:15 dorion: I think the problem is they're aiming for the masses rather than elite. think bitcoin can/will/is for liberating poor. so they push fast-food technology.

spyked: http://logs.ossasepia.com/log/ossasepia/2019-10-31#1008349 <-- ftr, I also stumbled upon this: http://thetarpit.org/2019/a-guide-to-systematically-exploring-the-entrails-of-mp-wp-illustrated-using-some-weird-found-in-the-post-editor . wrapping the input field content in htmlentities did the trick for me, but otherwise I thought this is something particular to my environment

diana_coman: http://logs.ossasepia.com/log/ossasepia/2019-11-01#1008378 - what is a "draft patch"?

ossabot: Logged on 2019-11-01 00:37:56 jfw: I had a couple draft patches for this and other issues; I could publish next week but will defer to diana_coman on priorities.

diana_coman: such a nice #o log to wake up to!

diana_coman: spyked: fast-food tech seems to me very aptly put, really; think of it: shiny and BIG, cheap and available in 1001 varieties of the same thing, "attractive" if you don't look much/haven't seen much else, shitty content and abismal quality that will screw you up in the end; what's there not to fit, really.

diana_coman: whaack: well done there; for the moment I'd say get out first what you had for this week anyway

diana_coman: then we'll see for next week; if jfw already has an actual solution implemented, there's no need to duplicate the work, can just as well pick his apart :P

diana_coman: and then billymg will pick apart that picking apart so hopefully the end result is fully tested.

diana_coman: spyked: + there is that important part "you don't have to do it yourself!!"

diana_coman: make it out of whatever shit you find, pre-packaged solutions tech or food-wise.

diana_coman: dorion: why u not in #t ? add it to your autojoin and don't be silly.

spyked: diana_coman, I can quite see the "one size fits all"/"pre-packaged silver bullet" angle. I'm only mildly curious what this says about the code itself, e.g. I looked at the google-android code at one point and it was grossly overengineered. and I suppose that's also the case with code that's otherwise proven to be working under extreme conditions, such as the linux kernel.

spyked: imho useful to make the distinction between "complexity for fast consumption" and "complexity that gives a huge productivity boost"

ossabot: (trilema) 2018-06-05 mircea_popescu: "thetarpit blog scaffolding is a few kLoC of CL, but lacks *any* editor-side interface" << understand something : i publish more than anyone, and by anyone we don't mean solo operators, we mean whole fucking outfits. there's a reason for this. the fact that trilema is comfortable to me provides those last edges of extra productivity and intellectual leverage that convert exceptional performance into mindblowing perfor

diana_coman: possibly; at the other extreme, the copy/paste/autogen horrors are anything *other* than overengineering, lol.

diana_coman: spyked: heh, yes, the two types of complexity mirroring the two types of simplicity

ossabot: (trilema) 2019-10-31 diana_coman: there's all sorts of simple and not all of it boring; that there though would be the boring-simple, more like the simple remains of a cut-out than the simple path to the core of an intricacy.

diana_coman: but fast-food is not really about no-complexity at all, quite on the contrary it would seem to me that it's precisely a lot of complexity just of the wrong kind (it's industrial for a reason, after all)

diana_coman: whaack lobbes jfw is this all about putting markup in the text or is there more to it (if yes, what)?

ossabot: (trilema) 2019-11-01 diana_coman: http://logs.ossasepia.com/log/trilema/2019-11-01#1949009 - hm; now I'll have to see this in detail then.

diana_coman: whaack: also, do delete those test-trackbacks, no need to actually keep them in there.

diana_coman: and worse, they obscure actual comments.

whaack: diana_coman: good morning. i implemented a solution last night to prevent select-display from hacking up other tags. I will give it a quick few tests and then publish it and continue with my week's work, if that sounds acceptable.

diana_coman: whaack: might as well publish it if it's done anyway but as a rule, before implementing a solution give some thought and/or ask re what the perceived problem is

diana_coman: because here essentially this is the q now that I have to figure out from you, jfw, billymg, lobbes : what *exactly* is the problem you set out to solve and why is it a problem?

whaack: diana_coman: http://logs.ossasepia.com/log/ossasepia/2019-10-31#1008365 i think lobbes example is user-error. putting b=http is asking to break a link. however in my example, there is a legitamite reason why one wants to select the text "trilema" but cannot do it because it is the same text found inside the anchor tag

ossabot: Logged on 2019-10-31 21:49:56 whaack: lobbes: i am working on a solution to this http://logs.ossasepia.com/log/trilema/2019-10-31#1948993 . you don't want to put any string for the b query param that is going to match a string inside of an anchor tag. this problem occurs frequently because often the same text in a link you want to select is inside of the anchor tag: i.e. for the text 'mp writes at <a href="http://trilema.com">trilema</a>' if you try to s

diana_coman: whaack: write it up and publish it since you have it, there's no harm in that and I'll read it, don't worry.

whaack: kk

dorion: http://logs.ossasepia.com/log/ossasepia/2019-11-01#1008384 << and building off diana_coman's comment, I see at least a couple angles. the first is who the technology is designed for. it seems since the '80s there has been a major push, primarily by aapl and msft, for computers for the masses, i.e. lowest common denominator, 'over 1 billion served'.

ossabot: Logged on 2019-11-01 06:32:40 spyked: http://logs.ossasepia.com/log/ossasepia/2019-10-29#1008077 <-- lulz @ "fast-food technology". I'm wondering how this translates to software in technical terms. is it just the inflation of code, or what?

dorion: this aim lead to 'point and click' (visual/spacial) rather than command prompt (verbal) usage modality. 'it should be so easy grandma can use it,' etc. further, the incorrect labeling of learned behaviour as 'intuitive', which really means luser ought never have to think, read manual or even error messages.

ossabot: (trilema) 2016-09-14 asciilifeform: 'the only intuitive notation is the tit' or how did it go.

dorion: To compensate for low quality user, all sorts of complexity needs to be built in, which is one part of the code inflation. another secondary part, described by naggum iirc (but don't have link handy) is demand for ever more 'programmers'. excess demand lowers quality on the creation side combined with sillycon valley, 'move fast and break things' so we can 'scale' (nevermind profitably) and sell

dorion: to the suckers on the public markets.

diana_coman: 'point and click' aka click-and-cluck

ossabot: Logged on 2019-08-29 19:13:57 diana_coman: shrysr: different yes but not better; it's usually touch-screens that they throw at kids from nursery, yes; + "learning to program by click-and-cluck"

diana_coman: and all of it coming from the core cause of too much money distributed to idiots; it makes for perverse incentives and the inevitable result of products made for them.

dorion: lol at click-and-cluck

dorion: more technically, "I'll import entire library rather than re-implement the one function I 'need'". and then the 'library' imports libraries.

ossabot: (trilema) 2018-08-21 mircea_popescu: Mocky he has a point, "library" is oreilly-ism. before the free/open source struggle for power, it was rather a teaching tool.

dorion: which leaves us in dark modern ages where can't buy professional computer -- everything is 'sysco' -- as well documented in http://loper-os.org , the logs, etc.

dorion: http://logs.ossasepia.com/log/ossasepia/2019-11-01#1008397 << thanks for the nudge. I was idling in there with a couple comments. there was a netsplit shortly after the new voice model spec and went back to reading logotrons and joining here rather than idel there.

ossabot: Logged on 2019-11-01 07:34:02 diana_coman: dorion: why u not in #t ? add it to your autojoin and don't be silly.

ossabot: (trilema) 2019-09-14 mircea_popescu: unrelatedly : hey trinque, now that there's actually multiple functional castles the time's prolly come to update the deedbot voice model into awareness of this situation. so how about a patch making voice in #trilema dependent on ~my~ wot, rather than deedbot's own ; and similarily in any castles that ask for it / you come to an understanding with the lords thereof, so they can use the voice model there if they wanna

diana_coman: dorion: meanwhile though I rated you and therefore you are not exactly idling in there; while you don't yet need to say anything, it's still better for you to be there.

diana_coman: for that matter hm, whaack , u also not in #t ?

whaack: diana_coman: nope i am not, I will change that now

dorion: diana_coman right, letting go of the old habits, aligning to present.

diana_coman: good.

diana_coman: sorry people for the spew from feedbot.

whaack: diana_coman: I rated you in a pm to deedbot, sorry for the delay. The rating was !!rate diana_coman 9 My Master The Marquess Eulora, may I demonstrate gratitude through my actions

diana_coman: whaack: no problem.

diana_coman: ftr the feedbot issue seems to be a flaw in how the bot decides which links are "new"; uncovered via unplanned testing when I increased the size of posts per page to make archives easier to use.

ossabot: (trilema) 2019-11-01 diana_coman: just 1. took out content from archives.php 2. increased limit at per page

jfw: Hi all, will catch up on log shortly. diana_coman: I've pushed through and got an outline. (Might be a bit verbose, for an outline.) Will see if I can hammer out the draft this evening.

jfw: http://logs.ossasepia.com/log/ossasepia/2019-11-01#1008388 - vpatch I made to capture my local change, not signed as yet.

ossabot: Logged on 2019-11-01 07:19:44 diana_coman: http://logs.ossasepia.com/log/ossasepia/2019-11-01#1008378 - what is a "draft patch"?

jfw: http://logs.ossasepia.com/log/ossasepia/2019-11-01#1008405 + http://logs.ossasepia.com/log/trilema/2019-11-01#1949028 - I expect I'll have to figure out the markup in text point before I'll see what's stupid about using markup syntax to do markup. Is one not supposed to have tags at all in posts?

ossabot: Logged on 2019-11-01 08:41:31 diana_coman: whaack lobbes jfw is this all about putting markup in the text or is there more to it (if yes, what)?

ossabot: (trilema) 2019-11-01 mp_en_viaje: i didn't look into it ; but going by the novel symbol <fn>, i expect he did a stupid thing, equivalent to walking on a rake, by declaring the footnote separators to be tag-like, <fn> and </fn>

jfw: How else does one do links if not <a href="...">? Images if not <img src="...">? And so on. There was the TinyMCE visual editor that billymg's patches removed, is that the preferred thing?

jfw: Then if I want to show the literal characters "", how's that supposed to work? The exact problem I perceived myself to hit here is illustrated as: if you use entities ala save, reload, save again, it turns into , which for one thing wasn't the text I saved and for another it then goes through to the browser as a tag.

jfw: (perhaps deeper: is there a WordPress manual somewhere that I missed?)

whaack: jfw diana_coman: I think that we may be mixing up a few separate issues. Issue (1) is the select-display, i gather this is what MP is referring to and he is saying don't put html markup in your b&e values. Issue (2) is the text editor for mpwp/browser evaluating the output of html entities

jfw: oh oh, I think you're right.

whaack: jfw: mp says don't do footnotes with <fn> </fn>, not that you shouldn't use any html tags at all

jfw: The selection bugs / operator errors are not the thing that I looked deeper at

ossabot: (trilema) 2019-11-01 diana_coman: http://logs.ossasepia.com/log/trilema/2019-10-31#1948998 - and moreover, jfw already looked deeper at it, so they'll coordinate and it will be fixed.

whaack: jfw: any chance that this http://ztkfg.com/category/a-sip-of-sicp/?b=Do&end=footnotes#select influenced your decision to do <fn> </fn> footnotes?

jfw: whaack: it is possible as I recall seeing the article before, though it's the sorta thing I might have considered anyway, don't recall exactly.

whaack: diana_coman: I answered you incorrectly for http://logs.ossasepia.com/log/ossasepia/2019-11-01#1008411 . The issue that the three of us were talking about was html entities evaluating in the text editor that I first mentioned here http://logs.ossasepia.com/log/ossasepia/2019-10-31#1008349 . And I also see now that I forgot to include the example of the problem in my post (which was on the unrelated issue I was working on), I will a

whaack: dd it now as a comment.

ossabot: Logged on 2019-11-01 10:35:54 diana_coman: because here essentially this is the q now that I have to figure out from you, jfw, billymg, lobbes : what *exactly* is the problem you set out to solve and why is it a problem?

jfw: I'll ask belatedly then: why is <fn></fn> stupid? Just the extra typing? I had noticed the footnote plugin originally used <footnote></footnote> (and still supports it for compat!), which did overflow my threshold of comfort

jfw: My reasoning was that there's already the entity quoting mechanism for entering a literal < (which is how this ties into the technically separate question of the editor), but none for any other sort of delimiter.

billymg: jfw: html tags are fine and meant to be used. http://logs.ossasepia.com/log/ossasepia/2019-10-31#1008357 specifically describes a bug in preserving "escaped" html entities (for example, when you want to output "" in your post, rather than have it be interpreted as the html tag )

billymg: as for footnotes, i think it's just easier and more convenient to use the footnotes plugin instead of manually editing markup to produce the same output

billymg: http://logs.ossasepia.com/log/ossasepia/2019-11-01#1008377 << no reason in particular, it was just the first method i stumbled upon that seemed to achieve the desired result

ossabot: Logged on 2019-11-01 00:32:12 jfw: billymg, know a reason to prefer htmlentities to htmlspecialchars? fwiw I can't tell why the former even exists and the docs don't fully specify what it does

billymg: http://logs.ossasepia.com/log/ossasepia/2019-11-01#1008394 << yes, i could include a test for that in the next mp-wp-tests patch

ossabot: Logged on 2019-11-01 07:24:36 diana_coman: and then billymg will pick apart that picking apart so hopefully the end result is fully tested.

jfw: billymg: oh, I certainly like the footnotes plugin compared to manual tagging & cross-linking, ugh. The possibly-stupid thing I did was change my delimiters from (( )) to <fn> </fn> (then cut my hand on the scythe because the regex code didn't quote the /)

billymg: http://logs.ossasepia.com/log/ossasepia/2019-11-01#1008411 << i think there may be a few bugs being discussed currently, i was only commenting on the "escaped html entities are not preserved in subsequent edits" bug

billymg: jfw: ah, i see, i must have missed that part of the thread

whaack bbl

diana_coman: http://logs.ossasepia.com/log/ossasepia/2019-11-01#1008471 - this, yes; can you tell me what problem(s) were you solving exactly? it got all muddied trying to follow what you were actually doing/talking about there.

ossabot: Logged on 2019-11-01 19:26:42 jfw: billymg: oh, I certainly like the footnotes plugin compared to manual tagging & cross-linking, ugh. The possibly-stupid thing I did was change my delimiters from (( )) to <fn> </fn> (then cut my hand on the scythe because the regex code didn't quote the /)

jfw: That if I used (( )), Lisp code would be a pain.

diana_coman: jfw: re &lt and the like, the question there is why would you want to write those in your text? (I suppose that's why I never got to see this)

diana_coman: jfw: I get that but just change that to some other delimiters, no?

jfw: You never needed to use a less-than sign, or illustrate a tag?

diana_coman: and yes, code is ~always in a pain in text, I agree.

diana_coman: I don't, no; give me an example where you have to.

diana_coman: (I know you *can* do it; I just haven't yet met the case where I *have to* do it)

billymg: diana_coman: can't speak for others but i ran into that when i was displaying some html code in a post http://billymg.com/2019/08/mp-wp-automated-testing-proposal-and-vpatch/

jfw: I have not met the case where I have to either; but - it's my blog, I don't want it telling me I can't, neh?

billymg: this line: "const postBody = 'When in the course of human events..."

diana_coman: billymg: what's the ip for that as I think I have only your old pizarro one

billymg: and without doing the whole < dance it would render the rather than just the text ""

diana_coman: jfw: heh, sure; the point is: why do you want to input &lt and the like?

billymg: diana_coman: 192.187.114.50

billymg: wait, no, that's BingoBoingo's box that it lives on but he provided me with another for my dns record. one sec...

diana_coman: billymg: hm? seems to work, lol

billymg: oh, weird

diana_coman looks

billymg: and yeah, it's the correct IP

billymg: but when i enter it into my browser it goes to http://bingology.net

diana_coman: jfw: btw did you read about feedback ? don't recall if I yet pointed you to it.

jfw: oh hey, I did already have a case of less-thans, in the line of http://fixpoint.welshcomputing.com/2019/misadventures-in-mp-wp-setup-the-sad-work-in-progress-post/?b=I%20guess%20the%20start&e=block#select

diana_coman: billymg: so inside <code> you mean?

billymg: yes, correct

diana_coman: jfw: do you usually write with &lt and the like or how come?

BingoBoingo: billymg: The bare ip in the address bar doesn't give apache enough information to know it should serve your site. Gotta send a name with the request

diana_coman: billymg: I think I tend to <cite>/<blockquote> code though yes, code is ~always a pain in posts/text (and I hate it too, because it IS text ffs, but can't say I have a solution that I'm happy with, really)

jfw: I did read the d-k rec thread and linked article when first posted, certainly some of the meaning escaped me but yes, what appears a problem might be the world giving a clue that you're being stupid

jfw: "do you usually write with &lt and the like" - no, and I haven't done that much writing in html really; if I end up with < in plain text and need to post in html, I can run it through 'sed' to escape them

billymg: diana_coman: yeah, it's definitely a pain and aside from typing out the entity code per character i don't have a better solution either

jfw: but was this a separate question from why change the footnote delimiter?

diana_coman: jfw: hence why it's always worth at the very least to first consider clearly what the proposed problem is; I'm still not entirely clear here as to what the problem(s?) you 4 suggest are exactly.

diana_coman: jfw: the main idea is that 1. the software should make it easy for the user to do what *they want* 2. including idiotic stuff if they are idiots

diana_coman: that's I suppose the main surprise in a sense: the trouble with protecting user against own idiocy is that the protection might protect idiots but it gets in the way of non-idiots most of the time

diana_coman: ie working with the user it's certainly preferable to write (( )) (or whatever delimiters you prefer eg ^^ ^^) rather than <fn> </fn>

diana_coman: and writing &lg is also not exactly what I'd ever want as a user, really

diana_coman: that being said, on one hand on your very own blog you can of course do exactly and entirely *whatever you want*

diana_coman: and on the other hand, at least from my point of view, different people can well have different preferences, sure

jfw: Needing < to input a less-than is not my doing, it's html syntax, if that was unclear

diana_coman: but I'd say it's at the *very least* worth giving a good thought to the alternative.

diana_coman: jfw: no, I know.

diana_coman: let me see now if I can in the end identify the various problems proposed:

diana_coman: http://logs.ossasepia.com/log/ossasepia/2019-11-01#1008478 - clash of (()) delimiters with Lisp, 1.

ossabot: Logged on 2019-11-01 19:45:00 jfw: That if I used (( )), Lisp code would be a pain.

diana_coman: http://logs.ossasepia.com/log/ossasepia/2019-11-01#1008485 - html code is a pain to display, 2.

ossabot: Logged on 2019-11-01 19:47:07 billymg: diana_coman: can't speak for others but i ran into that when i was displaying some html code in a post http://billymg.com/2019/08/mp-wp-automated-testing-proposal-and-vpatch/

diana_coman: jfw: re injecting html, that hm, if someone can plug whatever code into your *editor* page, it's already too late to protect against injection really.

diana_coman: whaack's problem selecting elements inside an html tag with the b&e mechanism, 3.

diana_coman: did I miss anything? jfw billymg whaack ?

diana_coman: lobbes: ^ ?

lobbes was about to respond to earlier ping, but that's all the info I have

lobbes: re: #3 that is

jfw: 2 was not among the issues I brought up, fwiw. And yes, the editor thing appears to be more involved than you've described there

jfw: (it's the only one I see as a bug really)

lobbes: (only thing I encountered was #3, and even that only last night. I have 0 knowledge of 1, 2)

diana_coman: jfw: help me here: what is the bug you see?

jfw: ok, so you want to display literal html code ala . You write it as , yes it's a pain, but it works

jfw: But if you save a draft and redisplay the editor, it shows up in the text area as - not what you typed.

jfw: clearer?

diana_coman: jfw: doesn't it work within <cite> or <blockquote> ? because code anyway doesn't quite work as plain text/outside of this ie it's not *just* those bits.

diana_coman: jfw: I suppose it's one of those: choose your pain; just choose wisely there.

jfw: It would be the same inside any other tag as I understand it

diana_coman goes in search of a post-with-code

diana_coman: damn it all, easier to test directly.

diana_coman: jfw: just tested here and inside <blockquote> it shows as &lt fine?

diana_coman: lettuce see the rest

jfw: did you navigate out then back in to the editor? just clicking save might not trigger a re-render

diana_coman goes to try it

diana_coman: jfw: hm, I did and still can't see it; mind putting a paste on paste.deedbot with an example so I know I'm testing exactly what you're having trouble with there

jfw: will do, can I test with a draft on younghands (as mine is patched)?

diana_coman: jfw: sure, why not

billymg: diana_coman: steps to reproduce: 1) create a new post, include some encoded html character, e.g. < 2) save post, 3) reload post edit page, 4) the < you previously saved will render as a < in the editor text area, so a subsequent save inserts the < into the db, not the < you had before

diana_coman: billymg: did that and...nope?

billymg: damn

diana_coman: but indeed, inside <cite> or <blockquote> or <code>, not just plain

diana_coman: why would you want it inside plain text since it's ..not?

billymg: yeah, same for me, inside code

billymg goes to try again

billymg: inside <code>*

diana_coman: I've put all 3 since I'm at it,lolz

diana_coman: ie one of each

diana_coman: anyways, I'll wait for pastes from you 2 and I'll try them

billymg: diana_coman: ok, i loaded up http://billymg.com/2019/08/mp-wp-automated-testing-proposal-and-vpatch/ in the mp-wp edit screen and indeed the text area contained "" instead of my originally saved ""

billymg: so for you when you navigate away / hard reload it preserves the "" ?

jfw: diana_coman: http://paste.deedbot.org/?id=DpUR

billymg: ^ that is consistent with what i am seeing as well

diana_coman: jfw: does that happen on younghands.club?

jfw: yes.

jfw: is it possible you did &lt without semicolon like you typed here?

diana_coman: billymg: I start suspecting it's something you broke with the latest /trimming patches really; iirc on ossasepia I have the original mpwp, ie not pressed to last vpatch

diana_coman: and that might be

diana_coman: jfw: no, I pasted exactly what you gave me there and it stays as you gave it

jfw: interesting! I also have the trimming patches

billymg: aha

billymg: ok, let me go look at the patch and see if i accidentally ripped out a htmlentities() or htmlspecialchars() call somewhere

billymg: ah, yes, found it

diana_coman: heh

billymg: "@param bool $richedit Whether or not the $content should pass through htmlspecialchars(). Default false."

billymg: although it's not immediately apparent to me why you wouldn't want $content passed through htmlspecialchars() regardless

billymg: diana_coman: before i spin up another local copy from the previous patch to test myself, was it in the html editor box that you tested behavior just now, or the "rich" editor box?

diana_coman: billymg: html editor box

billymg: ok cool

billymg: based on what i'm seeing it's hopefully just a flag to set to true somewhere

billymg: but will need to dig a bit more

billymg: sorry all!

diana_coman: jfw: thank you.

jfw: yw! so neither of us were entirely crazy.

diana_coman: now let's see, of the original 1,2,3 problems, is there anything left?

diana_coman: 1 is a matter of delimiters; 2 is apparently not in the original mpwp; 3 is a matter of citing properly the context rather than term inside link.

diana_coman: did I miss anything?

diana_coman: jfw: usually nobody's entirely crazy *to start with*; it takes a while :P

jfw: I am satisfied with using [[ ]] or something for footnotes because MP says so, though perhaps not entirely schooled on what's dumb about <fn>

diana_coman: jfw: essentially it's more you molding to the machine than the other way around really.

diana_coman: billymg: not a problem, happy we uncovered it at least.

billymg: yeah, i appreciate all the "testing in production" it received

billymg: also i see the mistake now, it'll be a one-line fix

diana_coman: jfw: one of those days I wanted to ask you re machines but atm I gather you rather have to focus on the writing, isn't it?

jfw: quite.

← 2019-10-31 | 2019-11-02 → ↑