3 ms·

hystry: Give me your browser history, and I will give you an inbox for the web

5 points by akkartik 20y ago

akkartik 20y ago
My goal has been to focus on two use cases: 1. When I return to a popular story on reddit/slashdot/digg I would like to see highlighted the comments that were posted since I last visited it. (digg is only in theory; does it ever get interesting comments?) 2. When I post a comment on a blog it usually has very little traffic. I never bother returning to see if someone responded to my comment, and would like to be reminded of it only when that happens. [update] All comments most appreciated. I will be only online one hour everyday for the next few days(I wrote this while on vacation travellling in india), so expect responses in 24-36 hours. Thanks.
staunch 20y ago
Give you my browser history, and I've lost my privacy and security. This is interesting, but unless you keep my history private I would never use this. It is possible for Firefox extensions to store data locally, so maybe you could.
- akkartik 20y ago
  This is a FAQ. http://hystry.com/faq.html http://hystry.com/faq.html Firefox extensions are restricted to javascript. Consider an extension that is about comparable in terms of computing resources: searching in scrapbook (http://amb.vis.ne.jp/mozilla/scrapbook http://amb.vis.ne.jp/mozilla/scrapbook ). On any reasonable set of pages saved, searching is dog slow. hystry's workflow - saving the state of every page you visit, and then performing diffs when you ask for them - is equally non-trivial. Also, scrapbook gradually makes firefox unresponsive as we keep adding pages to it. I have scrapbook setup to save every page I read, and need to save elsewhere the gigabyte of data it generates every month. This kind of computation and storage is what the website is for. hystry.com is pretty slow as well so far, but at least it doesn't keep popping up a dialog asking if you should let the javascript thread continue. And it will become faster over time.
  - staunch 20y ago
    For me the FAQ's answer is unsatisfactory. The claim that "The information cannot be connected with you" is a misleading falsehood. You can and so could any bad guy that has penetrated your system. No one who cares cares about their security or privacy could use your application in its current form. Thankfully there's lots of people who don't, so you should be fine. Good luck.
    - akkartik 20y ago
      Thanks. If you have any suggestions on how to make it work locally I'd be interested to hear them. Moving from a desktop to a web service will always involve some loss of privacy. The trend seems to be to deemphasize privacy.
      - ralph 20y ago
        A proxy server running on the user's machine? It gets to see everywhere the user goes. A Javascript or plugin can communicate with the proxy for better browser integration.
        akkartik 20y ago
        Hmm, I'll think about this one. Proxies are so far complex to install, perhaps this can be streamlined. I end up backing up an ever-growing scrapbook archive out of my PC every month. Perhaps most people don't care about backups. With hard disk capacities growing as they are, 100MB/month compressed may be a reasonable rate of disk usage.
        ralph 20y ago
        Be like Google and don't cache rich media in the proxy, just text/html, etc. An upstream proxy can be a full caching one, e.g. like Squid, perhaps the one at their ISP.
        akkartik 20y ago
akkartik 20y ago
jey 20y ago
Hystry is not something I'd use, but the demo impressed me. It seems to work simply and as advertised with no hassle and no frills. It does miss some changes though, e.g. on reddit a bunch of parent comments were listed as having been modified but the child comments were not. This seems like it would be better implemented entirely client-side. That way the privacy concerns automatically disappear, it'll be more responsive, and your server won't have to expend a lot of bandwidth/CPU/disk to fetch, create and store all the diffs.
- akkartik 20y ago
  Ah, so you've tried it! The html diff was harder than I expected (it has to handle comment permutations, for example) so I'm not surprised it has bugs. Can you point me to a specific URL, ideally one of the hystry.com/diff/ URLs? Much appreciated.

5 points by akkartik 20y ago

My goal has been to focus on two use cases: 1. When I return to a popular story on reddit/slashdot/digg I would like to see highlighted the comments that were posted since I last visited it. (digg is only in theory; does it ever get interesting comments?) 2. When I post a comment on a blog it usually has very little traffic. I never bother returning to see if someone responded to my comment, and would like to be reminded of it only when that happens. [update] All comments most appreciated. I will be only online one hour everyday for the next few days(I wrote this while on vacation travellling in india), so expect responses in 24-36 hours. Thanks.

staunch 20y ago

Give you my browser history, and I've lost my privacy and security. This is interesting, but unless you keep my history private I would never use this. It is possible for Firefox extensions to store data locally, so maybe you could.

akkartik 20y ago
This is a FAQ. http://hystry.com/faq.html http://hystry.com/faq.html Firefox extensions are restricted to javascript. Consider an extension that is about comparable in terms of computing resources: searching in scrapbook (http://amb.vis.ne.jp/mozilla/scrapbook http://amb.vis.ne.jp/mozilla/scrapbook ). On any reasonable set of pages saved, searching is dog slow. hystry's workflow - saving the state of every page you visit, and then performing diffs when you ask for them - is equally non-trivial. Also, scrapbook gradually makes firefox unresponsive as we keep adding pages to it. I have scrapbook setup to save every page I read, and need to save elsewhere the gigabyte of data it generates every month. This kind of computation and storage is what the website is for. hystry.com is pretty slow as well so far, but at least it doesn't keep popping up a dialog asking if you should let the javascript thread continue. And it will become faster over time.
- staunch 20y ago
  For me the FAQ's answer is unsatisfactory. The claim that "The information cannot be connected with you" is a misleading falsehood. You can and so could any bad guy that has penetrated your system. No one who cares cares about their security or privacy could use your application in its current form. Thankfully there's lots of people who don't, so you should be fine. Good luck.
  - akkartik 20y ago
    Thanks. If you have any suggestions on how to make it work locally I'd be interested to hear them. Moving from a desktop to a web service will always involve some loss of privacy. The trend seems to be to deemphasize privacy.
    - ralph 20y ago
      A proxy server running on the user's machine? It gets to see everywhere the user goes. A Javascript or plugin can communicate with the proxy for better browser integration.
      - akkartik 20y ago
        Hmm, I'll think about this one. Proxies are so far complex to install, perhaps this can be streamlined. I end up backing up an ever-growing scrapbook archive out of my PC every month. Perhaps most people don't care about backups. With hard disk capacities growing as they are, 100MB/month compressed may be a reasonable rate of disk usage.
        ralph 20y ago
        Be like Google and don't cache rich media in the proxy, just text/html, etc. An upstream proxy can be a full caching one, e.g. like Squid, perhaps the one at their ISP.
        akkartik 20y ago

akkartik 20y ago

jey 20y ago

Hystry is not something I'd use, but the demo impressed me. It seems to work simply and as advertised with no hassle and no frills. It does miss some changes though, e.g. on reddit a bunch of parent comments were listed as having been modified but the child comments were not. This seems like it would be better implemented entirely client-side. That way the privacy concerns automatically disappear, it'll be more responsive, and your server won't have to expend a lot of bandwidth/CPU/disk to fetch, create and store all the diffs.

akkartik 20y ago
Ah, so you've tried it! The html diff was harder than I expected (it has to handle comment permutations, for example) so I'm not surprised it has bugs. Can you point me to a specific URL, ideally one of the hystry.com/diff/ URLs? Much appreciated.