Support for Drupal 7 is ending on 5 January 2025—it’s time to migrate to Drupal 10! Learn about the many benefits of Drupal 10 and find migration tools in our resource center.
The recent and controversial break tag change have not included blogapi which still has the HTML comment style break tag.
We many resolutions and before raining this issue with patches, let's first decide which avenue we take
- We use a Drupal specific pseudo tag, hardwired. There seems to be little support for this -- even in the original issue I saw only two.
- We use a HTML comment tag, hardwired. Given one of the aforementioned two persons, the chances for .this is nil. But note that blogapi still has has tag hence this bug report.
- We use a variable without an interface which defaults to the Drupal specific pseudo tag.
- We use a variable without an interface which defaults to the HTML comment tag and we remove the update altogether.
- We use a variable without an interface which defaults to the HTML comment tag and we provide an update which reverts system_update_1018.
Code is written in http://drupal.org/node/87145 for the last three choices, so it's just a matter of decision.
This itme, let's wait for Dries decision and when he has made up his mind then we submit the patch from that issue.
Comment | File | Size | Author |
---|---|---|---|
#20 | rollback_0.patch | 3.38 KB | chx |
#9 | break_4.patch | 5.66 KB | Steven |
#4 | break_3.patch | 3.16 KB | Steven |
Comments
Comment #1
codepoet CreditAttribution: codepoet commentedI vote #4, FWIW.
Comment #2
andrewfn CreditAttribution: andrewfn commented#4 gets my vote
Comment #3
Steven CreditAttribution: Steven commentedUsability over backwards compatibility: the short tag should stay.
I also find this whole thing rather ironic. The thread where XHTML validity was the prime argument, got closed because someone screwed up the HTML.
And now, you want to change it because it was added by one core committer against the will of several people, yet propose even more complicated mechanisms to replace it. And then, you expect another core committer to make the decision.
My vote is to just fix the original bug: blogapi.module. Patch attached. I also removed a freak occurance of the break tag in the help text, which has no use as far as I can tell.
The break tag has been debated over and over, and has wasted too much time. We have more productive things to do.
Comment #4
Steven CreditAttribution: Steven commented*stupid losing attachment on preview bug*
Comment #5
m3avrck CreditAttribution: m3avrck commentedI agree with Steven, this patch looks good to me.
Comment #6
RobRoy CreditAttribution: RobRoy commentedWe write ".''. $node->body;" not ".''.$node->body;" :P I know it's in the original, but might as well fix it while we're in there. I would re-roll, but I'd be betraying my beliefs.
Even though there are some bad points in that thread, there are still some good ones. You/we should focus on refuting or confirming the VALID points for rolling this back if we actually want to get somewhere, not just bashing the poor arguments as there are quite a few.
Comment #7
RobRoy CreditAttribution: RobRoy commentedThere is just too much irony in this world, should have encoded that break tag. If only it was more usable... :D
Comment #8
eaton CreditAttribution: eaton commentedSince it was either missed or ignored, I'll simply post dman's comment from http://drupal.org/node/105278
I think it's pretty well established that there's no way the break patch will be rolled back or fixed. But it's worth explaining to dman why his work at data validation is not worth our consideration.
Comment #9
Steven CreditAttribution: Steven commentedAny direct use of the content in the node table requires post processing. Either you use the drupal filter system, or you roll your own. In the latter case, a simple str_replace is all that is needed. it is not the end of the world.
Here's a patch which should fix all bad code style regarding concatenation, given that you're playing the pedantic card, and still addresses the break tag bug.
Comment #10
chx CreditAttribution: chx commentedBefore others post patches for all five -- no need to reopen the debate. Now we have code for 1.) in this issue. Good. I am moving this to active because let's wait for Dries decision and when he has made up his mind then submit the patch. I do not want another 100 followup long thread with totally different patches being thrown in.
Comment #11
ontwerpwerk CreditAttribution: ontwerpwerk commentedoption 4 or 5 are my choices - because they:
<br>
and<break>
which would both be pronounced "break tag" on the phone to a userultimately I would say excerpt.module or something similar should be in core to make this whole discussion disappear
Comment #12
dman CreditAttribution: dman commentedThanks to Eaton for going in to bat and x-posting my rant.
I'm simply casting my vote and if I'm out-voted (or, um, over-ruled due to status) I'll live with it.
Yes, I can write a regexp just to deal with this one special case. And the next one, and the next one that the Drupal majority sees fit to inflict upon us. But I'll regard it as a failing of the Drupal data storage mechanism that it requires me to do so each time, and simultaneously prevents me from using the right tools to do so. Even the classic HTML 'tidy' tool will now totally barf on this unrecognised tag, and that can deal with almost any other bad data and markup the net (or Microsoft) could ever throw at it.
... you gotta try migrating some legacy content through several consecutive flavour-of-the-year formats and platforms some time. Heaps of fun.
Everyone thinks their own justification for ignoring standards is valid, and the folk who learnt the hard way are pedantic. *sigh*
Discard all the work done 'till now on getting XHTML to finally work right because you want to add one unneccessary custom flourish of your own and invent the incompatable DrupalML? please...
Data format lock-in (proprietary encoding of my data) like you dictate here when a compatable, valid, better, open, formalized alternative (XHTML) is ignored is enough to make most folk drop a software package altogether.
Please lets not be different just to be awkward.
.dan.
Comment #13
dman CreditAttribution: dman commented... however, to get out of rant mode and try and be constructive again...
I vote 4 or 5, But why not provide an interface to this variable so that admins who wanted to could just replace this marker with anything of their own choosing? In their own language even.
If they wanted to deal with the consequences (or more specifically wanted the poor bugger who came 2 years after them to deal with the consequences) and added invalid HTML, so be it.
It may be a cop-out according to some UI guidelines ;-) but if you can't get a solution that pleases everyone, just make it configurable under advanced options :-B
.dan.
Comment #14
webchickThe reason we can't really provide an interface to such an option is because every single node in the database would need to be updated each time this was changed, which is extremely expensive. So it should really only be something you only do once, but if it's available for users to click on they probably will. ;P
http://drupal.org/node/107061 is an attempt at a compromise which both retains the original break comment and improves usability for end-users. See what you think.
Comment #15
dman CreditAttribution: dman commentedTrue 'nuff.
... so don't do that then ;-)
Bad things happen if you change your files directory on-the-fly. So don't do that.
Random things may happen if you change the site theme without testing. Admins should not mess with it unless they really want to.
Changing the thumbnail size of images half-way through a live site will produce inconsistant results. Don't do that!
This config is not that much different.
It's for first-set-up admins, not users.
It's a really advanced admin decision, and should not be messed with... more than once.
Granted, It may be expensive to do a global search+replace on all nodes once (if that added option were even provided), but it's an admin function, not something that happens regularly.
You have a point, but not a great argument.
Comment #16
chx CreditAttribution: chx commentedhttp://drupal.org/node/107061
Comment #17
dman CreditAttribution: dman commentedI apologise for not reading to the bottom of the most recent http://drupal.org/node/87145 issue before posting above.
I'm happy with the default behaviour + optional super-admin (settings.php) override option (#4?) as seemed to be the outcome of that thread. Lets fry some bigger fish now eh?
Comment #18
dman CreditAttribution: dman commentedApologies for not totally reading each of the 5 strands this topic has now taken. The 'non-configuration' option of just setting it in the settings.php (of course, doh) is fine by me. I was reading the 'no interface' limitation a bit harshly. Sure, let it be set there ... at your own risk :-)
Comment #19
ericG CreditAttribution: ericG commentedprimary concerns should about usability as well as using standards.
usability for long term drupal users means not changing something so critical to their daily use of their site.
usability for site admins means not melting their database servers on an upgrade modifying node content needlessly
usability for developers means adherence to current web standards
all that means my vote is for #2, but am willing to see #4 as a valid compromise
If you want to do strange things to how the database stores menu info, url aliases, or anything that is truly drupal specific, then fine I have no problem with that, but changing the node content of people's sites to appease one stouborn developer? that is unacceptable. Do what you want with the rest of the db, leave my user-entered content alone!
Comment #20
chx CreditAttribution: chx commentedhttp://lists.drupal.org/archives/development/2007-01/msg00159.html states that Steven's JS won't go into 5.x and asks for choosing between options one or two and a clean up.
But there is a new argument on the field: blogapi.
Our support for blog APIs mandate a roll back. If I have a client which does a massive post to several sytems at once and the client supports Drupal then it could include an HTML comment to create a teaser for Drupal but it can not really insert a pseudotag as that would immediately break other system. I do not believe that forcing such clients to do extensive work to find whether the other end is a Drupal and insert the tag only if it finds a Drupal is acceptable. It's known to be rather hard to fingerprint a Drupal system if the administrator chooses to conceal this fact.
So the only acceptable solution is to roll back.
Comment #21
Caleb G2 CreditAttribution: Caleb G2 commentedThere is widespread agreement that either option, a) changing the break-command OR b) rolling it back is a temporary situation. The teaser has already been targeted to major change in Drupal 6 - this much is known.
So common sense is to use which ever temporary solution is less trouble to deal with while we work on fixing it the right way. Obviously not changing the way the break command has been worked for years is easier than changing it to something else for one release.
Lets please either: a) take the time to do it right NOW and be happy about it, or b) if that's not an option wait for calmer times (post 5.0 release) and do it right later.
Comment #22
Paul Natsuo Kishimoto CreditAttribution: Paul Natsuo Kishimoto commentedStandards incompliance also breaks future and present compatibility. Correct
and the nonsensical result is:
Compatibility and usability are inseparable; and you can't advance one by removing the other.
Voting #4, because a non-interface variable enables site admins to supplant
<!--break-->
if they feel the extra five characters will clog their databases and 'lopsidedness' will cause users to claw out their eyeballs. Those with 'more productive things to do' can leave<!--break-->
and immediately hook their valid XHTML content up to non-Drupal APIs.Comment #23
ericG CreditAttribution: ericG commentedre #20
When I read that post on the mailing list I read it as just another developer giving their opinion. I did not realize that it was the announcement of a decision and narrowing of the options. Thanks for being more explicit on what was between the lines of that post.
In that case, I'll just vote for a rollback to hardcoded as a comment tag (#2, although #4 is still a fine compromise), and go back to working on patches I've promised to some module maintainers.
Comment #24
webchickThis was rolled back in CVS.
Comment #25
webchickAnd closing this, to put this nasty stuff behind us.
Please see http://drupal.org/node/107061 if you're interested in making teasers kick-ass for 6.x!