Save Time On Hacker News: Mark All Read | Hacker News Reader

Save Time On Hacker News: Mark All Read

blog.hashtagify.me

100 pointsdanmaz7459 comments

I wish that when I read a large discussion (100+ comments) there would be a way to distinguish comments I have already seen. Because for a large discussion I care about, I'll revisit it several times to see what was added. Except right now, I have to rescan everythng.

When I was using micromobs.com for discussion groups, they had a very simple and elegant system: keep all comments in place to preserve context, but grey out the ones I have already seen. The problem on HN is that it would conflict with bad comments. Maybe use a different color?

For that I use this very nice extension: https://chrome.google.com/webstore/detail/fpndmkcfggkffpablc...

I use the HNHalfLife[1] userscript, and I know find it extremely annoying to browse HN without it. Thanks polyfractal!

[1]: https://github.com/polyfractal/HNHalfLife

I've just installed it (Chrome) but it doesn't seem to do anything, would you happen to know what I'm doing wrong?

Edit: Using the https version of news.ycombinator.com

I don't know, it "just worked" here. Click on comments page, close it and then it'll mark new comments when you return to it or see it on the stories list.

The hckrnews.com browser extension does this with an orange bar on the left of new comments, and on the hckrnews.com homepage it will list how many new comments have been added to the story alongside the total count.

You could try out: hncomment watcher http://christopherdebeer.github.com/hnCommentWatcher/

Yes, please, a way to see new comments in a discussion. One of the unfortunate side-effects of the current system is that interesting conversations get killed off too soon because nobody can follow what's being said any more. (Falling off the front page is also a killer for discussions. It would be nice if active discussion threads remained visible somewhere.)

I'm not the first person to bring up Usenet here, but it pains me that Usenet had way more functionality in the 1980s. (Which I don't mean as a criticism of HN, but a wish for more learning from the past.)

But how many users were there on usenet? The special thing about HN is that it keeps a quality of comments way above average without needing to be too elitist; I think that limiting the conversations (and thus the flame wars etc.) is a choice.

But to highlight new comments there are many different solutions already, take a look at the other comments :)

The special thing about HN is that it keeps a quality of comments way above average

That was in fact also a quality of Usenet, before it went out of fashion for unclear reasons.

It would be an interesting exercise to mirror (or even move) HN to a nntp-server. Personally I believe the signal/noise ratio would improve drastically.

I'd also be keen for a way to collapse the comment tree as I lose interest in responses to a particular reply.

I made something like that a while back: http://dionidium.com/projects/greasemonkey/markasread.user.j...

Cool. I expect kill-files will come afterwards. Several more extensions and hacks like that and we will have built a poor-man's USENET in the browser.

Well the original USENET was killed by it's lack of web features.

And file sharers, and spammers...

...just like we've built a poor-man's everything else in the browser. I cringe every time I visit a website with YA (broken) reinvention of YA GUI widget that has been polished and clean on desktop GUIs for two decades. (I cringe a lot.)

Great addition, but why not write it as a userscript instead of Chrome plugin? Chrome, Opera support them natively. You can install it just like an extension. Firefox needs Greasemonkey to be installed but most power users have it anyway. Not sure about IE and Safari but problably they can support userscripts in some way also.

This was just an experiment, and the extension looked like the easiest way to do it (I use chrome 90% of the time).

But I was thinking about publishing the source to github; the js is very simple, if anybody wants to fork it for other browsers is welcome :)

By the way, can you use css files with userscripts?

You can but they need to be hosted somewhere or you need to create a link tag with javascript and add it dynamically...

Something like that:

var css = document.createElement('link'); css.textContent = 'css here'; document.head.appendChild(css);

This is what I was afraid of. By the way, without the css the functionality works anywya, it just looks worse.

If the project is on github, you can just link to the repo for hosting.

I wrote something like this as a user script that works in FF(w/Greasemonkey)/Chrome/Opera a while ago [1]. The idea is to remove individual stories or add filters rather than removing everything like the OP. My script is based on VI-style shortcut keys though so that may influence how useful you find it.

[1] https://github.com/dnewcome/hn-plonk

Definitely a cool idea but it needs some work.

* Upon installing the plugin, I tried to "hide read" thinking it would hide the posts that I had already viewed, but they were not hidden. It should be possible to hide these with a css selector.

* After installing and clicking a post to read it, then clicking hide read, this post was not hidden (must not be considered read, again, css selector)

Finally I have a feature request: I would like a "mark read" link to appear next to each link, so that I can mark an individual link read without reading it.

Great idea and I hope that my feedback can help you to improve the app.

Thanks.

Thanks for the feedback.

It could be possible to mark individual items read, but hiding the ones you actually clicked on (without marking them read) is unfortunately impossible: As pointed out somewhere else in the comments, with javascript you can't read the "visited" selector, and with only css I could perhaps hide just the visited link, but not the encasing and trailing TR.

I figured out another way you can hide links as they are clicked.

With jQuery override the .click() event of every link on the page. Then, when the link is clicked get the URL of the link from the event parameter that is passed to .click(). From that information, you can construct a CSS selector like this one (with jQuery)

$("a[href=\"http://stackoverflow.com/questions/10668292/is-there-a-setti...)

Which can then be used to hide the clicked link. The problem now is one of persisting the data between sessions. Your options are localStorage, remote storage, and my suggestion: sqlite in webkit a'la: http://stackoverflow.com/questions/6200976/sqlite-in-chrome

After you have saved the "read" urls in SQLite, you should be able to hide the surrounding data as well in the other <tr> tag by traversing the DOM with jquery.

I really would love if you did this. Please consider giving it a try.

It looks like I should have done better markdown formatting on my selector example. For clarity:

    $("a[href=\"http://stackoverflow.com/questions/10668292/is-there-a-setting-on-google-analytics-to-suppress-use-of-cookies-for-users-who\"]")

this is extremely useful and it is one more reason why I really miss Chrome (and Firefox) on both tablets and smartphone. I read HN mostly on these devices where HN UX is really poor, this is one of those features I would love.

Another one is having bigger upvotes buttons and easier clickable links. Too often I have a hard time to click on the microscopic comments link, other times I open the wrong news or upvote the wrong one.

I find http://ihackernews.com/ to be pretty useful for small-screen devices.

I made a similar Greasemonkey user script a while back, and it's been great:

http://coding.pressbin.com/74/Update-on-Greasemonkey-script-...

Cool - so for FF there is already a solution :)

Looks cool. It would be great if you could differentiate between the items you've actually visited and those you've "marked as read."

I tried to do that with CSS :visited selector, but without success at first - I'll try again as soon as I'll have time!

Modern browser won't let you change :visited selector except color... because of security. A random webpage would have the knowledge of pages you have visited then :) I think IE7 and below have that bug. And you can't check the color of a visited link with javascript.

You were able to build a list of controversial webpages (to compare them with my browser history), put to your css file that a:visited color: red; (just an example) and check with javascript which links have foreground color red - and voilaa. You know what pages I have in my history.

You need to save data to localstorage or something like that instead. And you can only save those clicks which are made from hackernews site... not populate it from browser history.

I'll have to try with just css then. I can't keep track of all clicks :)

But you can't identify which links are visited... because of security issues I described. Sorry my english is horrible maybe I described it very in a very bad way.

https://bugzilla.mozilla.org/show_bug.cgi?id=147777

It is fixed in modern browsers now...

You can't with javascript, but the a:visited selector works in pure CSS, doesn't it? I'd like to do it so that I don't know which links you actually visited, but you would see them in a different color anyway. I think this should work somehow, I'll have to make some more tests.

You can only change the color, nothing else... :)

And that's all I need ;)

Great addition, but not compatible with this: https://chrome.google.com/webstore/detail/lgoghlndihpmbbgmbp... Probably because of infinite scroll, I have to scroll down a little bit to load few pages, then select hide read ...

I just tried and it works (even if not perfectly, as the colors aren't the same). What is your problem exactly? Did you try the button at the top of the page instead of the one at the bottom?

Love it! . . . But . . . I have Chrome set to delete all cookies whenever I close my browser and to save session cookies otherwise. So this extension works as long as the browser hasn't been closed once. But after the browser is closed and cookies deleted, the extension forgets what has been read when HN is reopened.

Actually it uses html5 localstorage, but if you delete that too, the list gets reset...

I wish there was more than one rss feed. What I really want to appear in my rss reader are the highest rated stories, not all of them.

As it is, using the rss feed is actually more inefficient than visiting the page every now and then.

There's an account on twitter called @hackernewsbot that tweets the highest rated stories. Twitter accounts have RSS feeds, which are a bit awkward to find, but it's been automated, just type the username in here: http://charlieharvey.org.uk/page/twitter_rss

Which gives: https://api.twitter.com/1/statuses/user_timeline.rss?screen_...

Hopefully this gives what you were after?

Hmm, I did that without looking and it seems the RSS urls aren't complicated any more. They used to need your uid not your screen name, which is why that tool exists.

Perhaps http://feedhint.com/

Cool. Something I asked long time back. Thanks.

Cool, but not really compatible with HTTPS Everywhere :( Have to browse HN on HTTP to see the extension in action.

Uhm... this could be very easy to fix, if it works I'll update it soon.

Now it works with https. Just update the extension.

I read hacker news from thunderbird via RSS. This problem doesn't exist for me :)

Someone should consider implementing this functionality into the site.

Tried to install. Got an error. "Manifest file is invalid."

Sorry, it was probably while I was uploading a fixed version for https. Try again and let me know.

PS: it's really odd though, when I uploaded the new version it gave me an error and I didn't get to save the changes. I didn't expect the broken manifest to be live!

Worked now. Thanks!

shouldn't it toggle read/unread?

This Firefox extension does it:

https://addons.mozilla.org/en-US/firefox/addon/link-visitor-...

thanks. i haven't used firefox for 3 years though.

Works great, but how do I undo?

Right now you can't... what is your use case? Was it just because you were testing, or you would really need an undo?

Reply on news.ycombinator.com