search

datalad / datalad.org

Website sources of datalad.org
https://www.datalad.org
4 stars 10 forks source link

Please kill datalad/pelican-bootstrap3 #8

Closed aqw closed 6 years ago

aqw commented 7 years ago

@yarikoptic @mih datalad/pelican-bootstrap3 is no longer needed.

---Alex

mih commented 7 years ago

Let's wait a bit with that. We still have that annex benchmarking article to go somewhere. Once we figured that one out the old page is truly obsolete.

aqw commented 6 years ago

@yarikoptic Is there any other unique content hosted on old.datalad.org? I can easily just grab benchmarking page as is, and have pelican copy it into the website statically.

That, or it can be turned into a proper page in the new theme. I don't remember what the blocker was for that though.

---Alex

yarikoptic commented 6 years ago

I do not think there was anything other valuable. The whole or majority of that page is coming from an ipython notebook IIRC using some plugin to render it into pelican. But since one time show, indeed might be easiest just to get a static copy and be done

aqw commented 6 years ago

@yarikoptic Cool. Then I will do so tomorrow. And then both old.datalad.org and datalad/pelican-bootstrap3 can be retired. :-)

---Alex

aqw commented 6 years ago

Ok. I went a little overboard, and ended up converting the page to use the new site theme. Since you said it likely won't be re-generated, I took the time to clean up the generated HTML so it's readable by humans too. ;-)

While I was at it, I extracted the images (so they're no longer BASE64 encoded in the webpage), gave them a background (so their text is legible in webkit and blink browsers), and optimized them to reduce payload size.

Let me know if you find any problems I introduced, but I think it's all good.

And then, I think datalad/pelican-bootstrap3 can be retired. :-)

---Alex

yarikoptic commented 6 years ago

Thanks @aqw ! looks great. Should we have it referenced to somewhere on the website since it is there now in a representable shape? may be within Development page?

aqw commented 6 years ago

@yarikoptic Yeah, it would make sense to link to it /somewhere/. ;-)

I agree, the Development page makes most sense. Though that page, in general, needs to be rethought so that it more intuitively conveys the info we wish to share. Another link on it won't hurt though. :-)

As an aside, I'm curious how the FS tests would look using ZoL 0.7.x. There were huge performance gains (especially for metadata) with the 0.6->0.7 upgrade. "Someday" we should run those tests again. :-)

---Alex

yarikoptic commented 6 years ago

Yeah and In principle it should be possible without major pains. I would just need to buy a few new drives to test on and be careful to not kill existing drives with data ;-) all the scripts are here I think https://github.com/datalad/benchmarking

aqw commented 6 years ago

@yarikoptic The link has been added to the dev page. So all is well now. Thanks for the link to the benchmarking repo. I've spent to morning distracting myself, with assorted cleanups in some of the scripts. A PR will follow once they begin to make sense. :-)

yarikoptic commented 6 years ago

Ouch... Now your are doomed to run it! ;-)

aqw commented 6 years ago

@yarikoptic Haha. I am ok with that. But I would need some drives... ;-)

aqw commented 6 years ago

@yarikoptic Also, as a followup, any chance of retiring datalad/pelican-bootstrap3 and old.datalad.org so we can close this issue? I don't have cool-kid-admin privileges for the repo. ;-)

---Alex

yarikoptic commented 6 years ago

retiring datalad/pelican-bootstrap3 is just a matter of git rm I believe (for .gitmodules as well) as for old. -- powers are of @mih

aqw commented 6 years ago

Sorry, I'm a bit confused. I can delete the "old" branch in this repo, and mih can retire the DNS, but the actual datalad/pelican-bootstrap3 repo, I don't have perms for. I assumed you did though.

---Alex

mih commented 6 years ago

Repo is gone.

yarikoptic commented 6 years ago

ah... it would've not hurt to keep it around "just in case" since it doesn't consume anything for us and neither is in the way of things, but that is ok

yarikoptic commented 6 years ago

@aqw re https://github.com/datalad/datalad.org/issues/8#issuecomment-408822974

aqw commented 6 years ago

@yarikoptic No PR yet, as the changes just kept going, and I didn't really have a setup to test them on. Not sure if you want expansive rewriting of the scripts with no testing. :-/

BTRFS has some cool potential, but the oops with their parity RAID levels left a pretty bad taste in my mouth. My "great hope for the future" is DragonflyBSD's Hammer2, although it's doubtful that it will ever be ported.