I2P dev meeting, December 18, 2012 @ 20:00 UTC

Quick recap

  • Present:

    Astral_12, eche|on, KillYourTV, LaughingBuddha, RN, str4d, topiltzin, zzz

  • Next Meeting

    The next meeting is scheduled for Tuesday, January 8 @ 20:00 UTC (8:00PM)

Volledige IRC Log

20:08:58  <RN> meeting start while I was disconnected?
20:09:17  <eche|on> which meeting?
20:09:38  * KillYourTV points to the topic
20:10:09  <RN>  Dev mtg here 8:00 PM (20:00) UTC Tues. Dec. 18^^ topic
20:11:38  <KillYourTV> maybe they've stopped...
20:12:12  <LaughingBuddha> Just start whenever?
20:18:08  <RN> usually dg runs the meetings
20:18:27  <RN> I think lillith took over when he had to leave one time
20:18:50  <str4d> He didn't show up last week, that's why.
20:18:53  <LaughingBuddha> Is there a link to the general agenda?
20:18:53  <str4d> (And the week before he was busy)
20:21:29  <RN> well, there is usually an agenda...
20:22:25  <str4d> There's a link to the agenda he had been using for the previous few meetings, but I can't pull zzz.i2p at the moment...
20:22:28  <RN> on trac
20:22:28  <eche|on> thats the update weather
20:22:28  <eche|on> hell a lot of updates/traffic currently
20:22:28  <RN> http://trac.i2p2.i2p/wiki/meeting
20:22:28  <str4d> o/ topiltzin
20:22:28  <eche|on> far faster than any update before
20:22:28  <str4d> RN, can you access that?
20:22:28  <Benedikt> what happened? one of the servers from this netword dropped down?
20:22:31  <str4d> I've got a few planned topics anyway, if you can't.
20:22:31  <RN> agenda hasn't been updated since the dec 4th meeting it seems
20:22:31  <topiltzin> \o str4d
20:22:31  <eche|on> Benedikt: net split, does happen a lot due to flaky i2p tunnels
20:22:31  <Benedikt> oh, thx
20:22:34  <str4d> Okay, I'm just going to propose a pseudo-random agenda.
20:22:45  <LaughingBuddha> Shoot
20:23:07  <str4d> (0) Say Hi.
20:23:07  <str4d> Hi!
20:23:15  <LaughingBuddha> Eyyy
20:23:31  <topiltzin> what up
20:25:46  <str4d> Any other takers? ^_^
20:25:52  <str4d> Alright, moving on:
20:25:52  <str4d> (1) Network Health / Update Status
20:26:31  <str4d> Network health seems (to me) to have been pretty bad the last few days/week or so. Do we know if this was an actual attack?
20:26:46  <str4d> Or just continued degradation due to the bugs?
20:27:58  <LaughingBuddha> I guess we won't know for sure until more people actually update to 0.9.4
20:28:29  <RN> str4d, the trac link? yes
20:30:43  <topiltzin> eche|on was reporting some weird behavior, but we would not know for sure even after people upgrade to 0.9.4
20:34:04  <str4d> psi seemed to think that some of it was related to the Russian guides recommending that users force floodfill to find more nodes.
20:34:07  <str4d> With regard to updates:
20:34:15  <str4d> <eche|on> hell a lot of updates/traffic currently
20:34:18  <str4d> <eche|on> far faster than any update before
20:34:29  <str4d> Anything else to add to this summary of update progress eche|on?
20:36:17  <eche|on> 780 full updates pushed till yet by my nodes
20:37:38  <str4d> LaughingBuddha said that torrent updates worked for him. Anyone know how large the swarm is?
20:37:54  <str4d> (Only dev routers I know, but would be interesting to know)
20:38:13  <KillYourTV> Just around 10
20:38:59  <eche|on> the torrent is very low usage for me
20:39:06  <KillYourTV> or, let's put it like this: I see 10 peers
20:39:17  <eche|on> I see 6 for su2 and 4 for sud
20:40:01  <KillYourTV> since only a very small percentage of dev build users were offered the torrent update, I'm not surprised by there being so few transfers thus far.
20:40:16  <Astral_12> str4d: 6-7 peers , su2
20:40:30  <str4d> KillYourTV, no, all dev build users are offered torrent.
20:41:09  <str4d> If version ends in -0 and you pick a number from 0 to 99 that isn't 0, you don't get torrents.
20:41:12  <str4d> (IIRC)
20:41:15  <KillYourTV> oh...i misunderstood then.
20:41:23  <KillYourTV> http://tracker2.postman.i2p/details.php?dllist=1&filelist=1&info_hash=%09k%a6%29%14%7b%15%f6%f89%d7%1b%1c%d9T%fe%60c%ec%7c
20:41:23  <iRelay> Torrent #19559: i2pupdate-0.9.4.su2, Size: 3.06 MB, Downloads: 6, Added: 2012-12-17 17:52:22, S/L: 10/1
20:42:01  <str4d> It's enough to show that the update mechanism is working though (and we don't need to worry much about scalability, because that's already been well-tested for torrents ^_^)
20:43:11  <iRelay> <ReturningNovice@kytv> I missed the torrent option
20:43:37  <iRelay> <ReturningNovice@kytv> was upgrading from 0.9.3-13
20:43:48  <str4d> Okay, not much more to say here for now - we can appraise the effects of the update at the next meeting.
20:43:48  <str4d> (2) Website Revamp
20:43:48  <str4d> http://vekw35szhzysfq7cwsly37coegsnb4rrsggy5k4wtasa6c34gy5a.b32.i2p/en/site/ is where I am currently up to (make sure to hard-refresh to get the latest CSS)
20:43:55  <iRelay> Title: The Invisible Internet Project - I2P (at vekw35szhzysfq7cwsly37coegsnb4rrsggy5k4wtasa6c34gy5a.b32.i2p)
20:44:27  <str4d> RN, your router probably found the update via HTTP first.
20:44:30  <eche|on> str4d: scalability is a problem with a non reachable torrent
20:44:37  <zzz> there is no option
20:44:40  <eche|on> and currently ONLY dev builds gets a torrent
20:44:59  <str4d> Check http://localhost:7657/debug - you should see a TORRENT checker there.
20:45:14  <zzz> and no, 20,000 in one swarm is not well-tested within i2p.
20:46:14  <eche|on> str4d: and I do not see a torrent on 0.9.3-15
20:46:21  <str4d> zzz, true, I'd forgotten that *derp*
20:46:32  <LaughingBuddha> str4d: Seeding: 0 / 9 peers
20:46:55  <str4d> zzz, how does the update manager choose between two update options when they have the same priority?
20:46:58  <eche|on> [RegisteredUpdater org.klomp.snark.UpdateHandler for ROUTER_SIGNED TORRENT @pri 10] *oooüs*
20:47:08  <str4d> Is it just whichever one is found first?
20:47:50  <zzz> dont know
20:48:39  <str4d> Looking at the paste above, I'm guessing that it's highest priority number wins rather than lowest.
20:50:15  <str4d> Does anyone have anything else they want to add here? We can appraise the effects of the update at the next meeting.
20:51:21  <topiltzin> site looks good here :)
20:51:28  * str4d has to leave in 25 mins so wants to move on
20:52:57  <RN> please continue str4d
20:53:21  <str4d> (2) Website Revamp
20:53:22  <str4d> I've tweaked the old themes so they at least refer to the right parts of the HTML, so the links in the footer can be used to try out the different CSS files available.
20:53:52  <str4d> It gives an idea of how the new structure can still be differently layed out.
20:55:17  <MTN11> i2p: kytv@mail.i2p * rb47286b470722a6382a963092219407946bcc5b6 installer/ (5 files in 5 dirs): Add dummy placeholder files to empty eepsite directories
20:55:17  <MTN11> git and a few other vcses don't store empty directories....but now these directories
20:55:17  <MTN11> are no longer empty.
20:55:32  <jenkins> Starting build #46 for job i2p (previous build: SUCCESS)
20:59:21  <jenkins> Project i2p build #46:SUCCESS in 2 min 42 sec: http://kzzj7cu24fo6w7jsisgysh65lfyrkjaaaqsdn3hvlop6lp36vs2q.b32.i2p/job/i2p/46/
20:59:24  <str4d> How does it feel to people?
20:59:24  <str4d> I've also made several backend changes that should hopefully make management of the site easier - e.g. incrementing router version only has to be done in two places now zzz
20:59:24  <str4d> (Well, really only one, but I added another one so I could add Launchpad to the mirrors list)
21:01:40  <str4d> Did anyone see my posts above?
21:03:08  <KillYourTV> yes. I think the site looks very good.
21:03:31  <LaughingBuddha> Yup
21:03:37  <LaughingBuddha> +1 on the design
21:03:37  <str4d> laaaaag
21:04:45  <KillYourTV> a few broken links though bu tmaybe those are known
21:06:05  <KillYourTV> specifically http://vekw35szhzysfq7cwsly37coegsnb4rrsggy5k4wtasa6c34gy5a.b32.i2p/en/site/volunteer/develop/developerskeys links to http://vekw35szhzysfq7cwsly37coegsnb4rrsggy5k4wtasa6c34gy5a.b32.i2p/en/site/volunteer/develop/monotone which doesn't exist.
21:06:55  <str4d> I still plan to open up the design to suggestion from the community (since there are still things about duck's design that bug me) but as a first iteration it works.
21:06:58  <str4d> KillYourTV, as per ticket #807 the broken links are known (mostly on the earlier pages I migrated, since at the time I didn't know where the other pages were going to be).
21:06:58  <str4d> Questions:
21:07:01  <str4d> - Are the Blog and Meetings feeds alright as they are, or should they truncate their content (so readers have to go to the main site to view everything)?
21:07:05  <iRelay> http://trac.i2p2.i2p/ticket/807 - (accepted enhancement) - Revamp of website
21:07:36  <str4d> - Does the Meetings feed need to be more obvious, or is the <link> on the meetings index enough (it gets picked up by Firefox under Bookmarks -> Subscribe to This Page)?
21:08:39  <str4d> KillYourTV, oh, that's a different bug - the page exists, but I shifted the monotonerc file it imports, so I need to fix that ><
21:09:07  <str4d> But again, all links like that will be fixed.
21:09:21  <str4d> - Does anyone have suggestions as to content changes that need to happen before the revamp goes live?
21:09:35  <str4d> - Do we need RSS feeds as well as ATOM?
21:10:55  <zzz> are all the pages staying at the same URLs?
21:10:58  <str4d> - Should the "any" link in mirrors preferentially choose a mirror in a country determined from the viewer's language?
21:11:44  <topiltzin> that's a very tricky one
21:12:30  <topiltzin> what is the current behavior?
21:13:08  <str4d> zzz, I'm reasonably happy with the urls as I have them now (but suggestions are welcome).
21:13:18  <str4d> There are a handful of pages I have been unable to place though.
21:13:21  <str4d> topiltzin, see http://vekw35szhzysfq7cwsly37coegsnb4rrsggy5k4wtasa6c34gy5a.b32.i2p/en/feed/blog/atom and http://vekw35szhzysfq7cwsly37coegsnb4rrsggy5k4wtasa6c34gy5a.b32.i2p/en/feed/meetings/atom for the current behaviour
21:13:27  <iRelay> Title: I2P Blog0.9.4 Release (at vekw35szhzysfq7cwsly37coegsnb4rrsggy5k4wtasa6c34gy5a.b32.i2p)
21:14:02  <zzz> str4d, I mean with existing site vs. new site
21:14:32  <str4d> zzz, if you were asking if the pages are at the same urls as the current site: no.
21:14:35  <str4d> The urls *actually make sense* now.
21:15:25  <str4d> zzz, there are a few legacy helpers which redirect old urls to new ones, but that only works for some of the pages.
21:15:32  <zzz> str4d, we need to talk about that then. We have a lot of SEO / link juice built up. There's also a couple under /static/ that have to stay the same
21:15:46  <zzz> hosts.txt subscription, and backup news
21:15:50  <str4d> (namely, meetings, status, and any pages placed into the legacy map)
21:16:04  <zzz> we also have a lot of links back to the website in our console
21:16:04  <str4d> zzz, the static ones are the same AFAICT
21:16:37  <zzz> can you have a redirect for every old page?
21:17:46  <str4d> LEGACY_MAP={
21:17:52  <str4d>     'download': 'downloads_list'
21:17:52  <str4d> }
21:17:52  <str4d> ^-- we can add mappings there from the old site to the new.
21:17:52  <str4d> zzz, it would have to be a manual redirect, but yes.
21:17:52  <topiltzin> does this link juice build-up apply across all our domains?
21:17:55  <str4d> (I'd want to shift the legacy-supporting code into a separate file though, because that would be looooooong)
21:17:58  <str4d> And if I tell it to use 301 redirects, the client sees the redirect as permanent (or is that 302?)
21:18:33  <topiltzin> could we perhaps try the new site on geti2p.net and keep the old one on i2p2.de or similar?
21:19:00  <str4d> topiltzin, AFAICT geti2p.net is the same site (just a second domain)
21:19:07  <str4d> (as in, same backend)
21:19:42  <zzz> this is not a little thing. We've built up a lot of link reputation in the last 5 years and we can't afford to throw it all away. It's extremely important to get this right.
21:19:42  <topiltzin> I'm trying to figure out some way to traffic-test it sooner rather than later
21:20:10  <str4d> zzz, okay, I'll go through and add every page I've moved to the LEGACY_MAP
21:20:17  <zzz> another thing - that both with and without .html suffix works.
21:21:03  <str4d> zzz, already done (one-line addition to the legacy mapper)
21:21:03  <str4d> Also, can people suggest new url locations for the following files? (I'm stumped)
21:21:03  <str4d> i2ptunnel_services.html  jbigi.html  manualwrapper.html  minwww.html  ports.html  ratestats.html
21:21:34  <zzz> there's also the rel=canonical stuff
21:22:19  <str4d> zzz, can you elaborate?
21:22:26  <zzz> we actually spent a fair amount of time on SEO stuff trying to get our juice back after i2p.net went away
21:22:41  <zzz> in _layout.html
21:23:20  <zzz> you want to point back to the main site so mirrors don't suck away your reputation.
21:23:27  <zzz> thats also why mirrors are always tagged nofollow
21:24:10  <zzz> so you're not giving your rep away
21:27:32  <topiltzin> would a mirror work for a traffic test?
21:28:25  <str4d> Okay, well we just have to ensure that those are carried over to the new global/layout.html
21:30:02  <str4d> (I'm guessing duck didn't do that)
21:32:53  <topiltzin> (would love to read some chat logs about the SEO stuff)
21:34:37  <zzz> str4d, if you haven't been thinking about SEO yet, time to dive in ASAP. Every word on the home page matters.
21:35:00  <zzz> Both on the current home page and on the duck version, the words are chosen very carefully.
21:36:58  <zzz> look for example at where we rank on google for 'garlic routing'. I wrote that page from scratch specifically to get page rank for that search term.
21:40:37  <str4d> I've been more focused on getting the new site operational. But I haven't changed much text content.
21:41:36  <str4d> I can go through and make sure the new layout dude doesn't lose any link stuff. Do you have suggestions for the SEO of the front page as it stands now though, zzz?
21:43:52  <str4d> Since the text content of the migrated pages has not changed, they should be fine once the legacy redirects are in.
21:45:20  <zzz> havent looked in months. whats the url again?
21:46:13  <zzz> also str4d, we should standardize on '-' instead of '_' in urls. Right now it's a mix, but in the last 10 years the web has clearly standardized on '-'. Underscores look bad and it doesnt matter for SEO any more
21:46:29  <str4d> It's above
21:46:32  <str4d> Also linked on ticket #807
21:47:19  <str4d> Yep, I noticed the mix. I'll fix that up.
21:47:22  <zzz> ok I'll check the ticket. all the links above I see are to feeds and such
21:48:19  <zzz> also good is to maintain short urls for the pages that get tweeted or referenced often - /download, /faq, etc.
21:48:22  <str4d> The links above-  use that b32
21:50:25  <zzz> and if you or somebody else trusted wants to go further and claim our site on google webmaster tools, add a site map, etc., that would be great. I'm not at all claiming our SEO is great, just that we used to think about it at least a little.
21:52:00  <str4d> I was going to add site map generation to the backend, so I'd be happy to do that.
21:52:18  <topiltzin> does google webmaster tools report data back to google?
21:52:58  <zzz> only if you put their js on your site
21:53:27  <str4d> Short URLs - would the legacy URLs be enough?
21:54:10  <str4d> topiltzin, only if we run Google Analytics
21:54:49  <topiltzin> but will we get meaningful information without reporting anything back to google?  or do we not care?  do paranoid users care? etc.
21:55:38  <zzz> sure, good questions. ofc we can also just awstats or similar across our logs looking at referers and search terms
21:55:45  <zzz> doubt anybody's done that in years.
21:57:16  <str4d> Well, the tools report back of course, but only because they are interfacing to Google data
21:57:26  <str4d> Does the webmaster need to be non- anon though?
21:58:06  <zzz> just need a gmail addy afaik
21:58:13  <str4d> topiltzin, we will get Google search info, which in itself is useful.
21:58:39  <str4d> Analytics is technically separate to the webmaster toolkit IIRC
21:58:46  <zzz> anyway, all this is a sideshow for now
21:59:21  <str4d> Yep. zzz, you reached the site yet?
21:59:44  <zzz> nope, been talking to you :)
22:00:49  <topiltzin> I like to pre-fetch eepsites while I chat :)
22:02:17  <zzz> don't have much extra upstream bw for fetching atm
22:02:20  <RN> " Umm... the server encountered some sort of error. "
22:03:08  <RN> that's what I'm getting when I select the FAQ from the dropdown menu
22:03:31  <zzz> str4d, how are you staying in sync with the old website? did you branch from it in mtn or are you doing it manually, or have deferred the whole painful issue?
22:03:34  <str4d> Rn, odd... shouldn't do that.
22:04:00  <str4d> zzz, branched in mtn and propping regularly.
22:04:23  <str4d> I'm changing no content in the revamp so merges work alright.
22:04:58  <str4d> There have been a few tricky ones, but mostly easy props.
22:10:49  <zzz> what are you asking for comments on ? broken links? general organization? layout/colors/CSS? I have no idea where you're at.
22:11:54  <RN> dark theme dosn't look right
22:12:00  <str4d> zzz, navigation layout, URL forms, general usability, and front page content
22:13:21  <str4d> RN, neither of the old ones do. I've asked dr|z3d for some pointers, but none of the current themes will likely be the final one.
22:13:51  <str4d> RN, re the error, I'll look at it in about 20 mins
22:14:07  <zzz> so there won't be any theme selectors in the final version, hopefully?
22:14:58  <zzz> multiple themes is a horrific waste of time
22:15:20  <str4d> zzz, probably not, unless there is indecision over the final theme. They are there to help designers.
22:15:43  <LaughingBuddha> They can always be created by users. It isn't something the devs should worry about
22:16:20  <LaughingBuddha> IMO
22:16:27  <str4d> I just put the links there to expose the functionality for now.
22:17:36  <zzz> the thing that bugs me the most about the home page is the black bar with the links. That aren't really links since they only popup a menu, they don't go anywhere ifyou click on them
22:18:37  <zzz> it's like it's the wrong color, and in the wrong place, and too small or something. It's not in the style of the rest of the page, and I'm wondering if all that should be below the green part
22:19:12  <str4d> zzz, I haven't focused on color or style.
22:19:49  <str4d> The decision to not link the menu openers came after usability discussions and feedback.
22:20:16  <str4d> It was decided that linking some of them and not others would cause confusion.
22:20:43  <str4d> The plan is to use CSS to make the distinction clearer.
22:21:35  <zzz> cant put my finger on it but the top 3 sections (yellow, black, green) seems like one too many or in the wrong order or something
22:21:35  <zzz> not just about color/style maybe? don't know. there's something unfinished or out-of-place about it
22:21:35  <zzz> maybe some styling can bring it all together, maybe not.
22:21:42  <str4d> The color and style of the navbar will be changed to integrate better, but I needed something functional.
22:22:00  <zzz> I'd love to find someplace on some page to put a big "Celebrating 10 years 2003-2013" banner
22:22:50  <str4d> (this is why I added the theme links, to show that thesaurus structure can be rendered differently)
22:23:24  <zzz> whats a thesaurus structure?
22:23:31  <zzz> havenc clicked on theme links yet
22:24:20  <str4d>  Argh, phone typing fail. should be "same"
22:25:06  <zzz> and now I wish I hadn't
22:26:09  <zzz> too bad dg seems to have gone up in smoke. All that youthful enthusiasm fizzled out so fast?
22:26:28  <zzz> two meetings and done
22:26:47  <str4d> yeah, half the old castle isn't being used properly because of the new structure. but I haven't had time to migrate them yet.
22:26:58  <str4d> css*
22:27:36  <str4d> It sounds like he's rather busy at present.
22:28:26  <zzz> so that's my takeaway. the black menu bar seems to be the most problematic. Not that I have any suggestions about it. Just feels/looks not right.
22:30:12  <topiltzin> I'm sure another dg will show up or the old one will return when the time is right :)
22:30:28  <str4d> Yep. what about the layout of links within it? Sensical?
22:30:28  <topiltzin> but in the meantime, we're still doing great and it's important to listen to non-youthful and non-enthusiastic points of view as well
22:31:10  <str4d> (also, I'll try changing the black bar to use the color scheme from ducks example nav bar.)
22:31:37  <zzz> just surprised to see dg go from 60 mph to zero in a couple days.
22:32:51  <zzz> dunno if we want direct links to the jar, exe, etc on the home page. Seems like the text on the download page is important.
22:33:45  <zzz> defnitely need a llnk to trac, looks like that got lost?
22:34:16  <zzz> if you would go somewhere when you clicked on the word, that would be a big help too
22:34:26  <KillYourTV> Volunteer -> Develop -> Bug Tracker
22:34:54  <zzz> hahgeez there's two levels?
22:35:32  <zzz> there's no little arrows indicating another level
22:36:09  <zzz> and do we really want two levels on the home page, or anywhere? might be more than topiltzin's grandma can handle
22:36:44  <zzz> and we darn well better send grandma somewhere when she clicks on download
22:37:43  <topiltzin> yeah lets make sure she can get i2p and worry about making her a developer later
22:38:15  <zzz> heck, might be best to scrap the fancy dropdowns and just replace download-about-help-volunteer with four big-ass buttons
22:39:01  <str4d> zzz, there will be.
22:39:23  <topiltzin> if the big-ass buttons are in Pretty Colors grandma would love it
22:39:34  <str4d> And it was topiltzin who suggested to move docs into the about menu, so there are techincally three levels
22:39:57  <str4d> zzz, I had the Trac link in the help menu, but topiltzin suggested moving it to voliunteer -> develop
22:40:11  <zzz> it's like we're trying to make a hierarchical website, but only thru a complicated multi-level menu bar on the home page. What we really need are actual hierarchical pages, like an about page, a volunteer page, and a help page.
22:40:57  <KillYourTV> Now 8 peers for the sud   http://tracker2.postman.i2p/details.php?dllist=1&filelist=1&info_hash=%f0%d8%27%17W%cfF%83%9em%9e%3f%d8%f8%85%2ac%baRV
22:40:57  <iRelay> Torrent #19558: i2pupdate-0.9.4.sud, Size: 7.09 MB, Downloads: 6, Added: 2012-12-17 17:49:15, S/L: 7/0
22:41:00  <zzz> trying to create strucrture solely in that black bar is insufficient, and maybe the wrong place to try
22:41:30  <str4d> zzz, the structure is in the urls. I copied that structure into the nav bar because it seemed to make sense to me.
22:42:13  <str4d> As I've said all along, the biggest task was deciding how to rearrange the old pages from a single directory into a hierarchy, and I have had almost no help there.
22:42:24  <zzz> just spitballing. Haven't looked at it in months and may have a different opinion tomorrow.
22:42:47  <zzz> not criticizing, just speculating
22:43:06  <str4d> So my biggest question is, do my structure decisions make sense?
22:43:25  <str4d> (i.e. to people other than me)
22:43:35  <topiltzin> it's good that we have so many future directions to explore, I'm sure it will result in many productive discussions
22:43:55  <str4d> Once the structure of the existing pages is set, then the other issues can be worked out around that.
22:43:58  <topiltzin> str4d: yes here
22:44:31  <str4d> And did anyone have suggestions to my earlier query about where to put the last few pages?
22:44:34  <str4d> i2ptunnel_services.html  jbigi.html  manualwrapper.html  minwww.html  ports.html  ratestats.html
22:45:00  <topiltzin> and since dg has proven that he can go from zero to 60 mph in a lot shorter time than few days, those discussions are going to be *FUN* </offtopic>
22:45:06  <zzz> what do you mean "put"?
22:45:09  <str4d> There are also older pages that I haven't added in yet, because I think they are somewhat irrelevant, but they should probably go *somewhere*:
22:45:12  <str4d> announcements.html  clt.html  cvs.html   i2ptunnel_migration.html  invisiblenet.html  myi2p.html        transition-guide.html  upgrade-0.6.1.30.html
22:45:12  <str4d> benchmarks.html     _config   feed.atom  installation.html         jrandom-awol.html  statusnotes.html  transition-guide.txt
22:45:16  <zzz> where to link to them?
22:45:23  <str4d> zzz, in the backend / the urls.
22:45:34  <str4d> All site pages are stored under pages/site/
22:45:45  <str4d> From there, the folder structure matches the urls.
22:46:04  <zzz> you're hierarcical-izing all the files too?
22:46:21  <str4d> So e.g. the page on unidirectional tunnels is stored in pages/site/docs/tunnels/unidirectional.html
22:46:36  <zzz> because that seems like a fools' errand to hyper-categorize everything
22:46:39  <str4d> zzz, yes - makes site management a hell of a lot easier both on the people and the code.
22:47:13  <zzz> if you say so. but for gosh sakes just make a misc/ directory and be done.
22:47:48  <zzz> remember a file can only be in one place but there can be multiple links to it. don't get yourself stuck trying to put everything in its place.
22:48:26  <str4d> Alright, misc it is. And I disagree there - at least from an SEO perspective, unique content should have a unique slug
22:49:00  <str4d> Hence, I wanted to try and get the urls right.
22:49:04  <zzz> i think if you just pulled out the release notes, meeting logs, and statuses into their own dirs  you'd find there wasnt a lot left
22:49:35  <zzz> sure, unique URL, but you can link to it from many places. Not every link from tunnels/index.html has to be to a tunnels/xxx.html file
22:50:16  <str4d> Yes, true (that itself is why half the links are still broken)
22:50:31  <zzz> e.g. on how.html there's dup links in there, and links to all sorts of topics.
22:50:34  <str4d> But I wanted the url of each page to make sense for that page.
22:51:05  <str4d> i.e. /en/site/docs/transport/ssu vs /en/site/docs/spec/ssu
22:51:16  <str4d> (There are both - one is the documentation, one is the specification)
22:51:46  <zzz> nice in theory, maybe not so much in practice. or maybe it's all good.
22:52:23  <str4d> But if everyone is happy enough to have an /en/site/misc/* url for the remaining pages, I'll put them there (as I'm getting tired of page sorting @_@)
22:54:09  <zzz> just don't try to achieve some mythical and useless goal of a 1:1:1:1 mapping of menus:topics:urls:filenames
22:55:07  <zzz> and a deep hierarchy doesn't help anybody. hurts SEO, hurts the people typing, hurts readability of the urls, etc.... so dont get carried away
22:55:10  <str4d> b
23:00:46  <str4d> With regard to that, I'm wondering if there is a way to remove "site/" from the url - it just adds to the length.
23:01:04  <str4d> But given that other pages are e.g. "/en/download" or "/en/blog", having "/en/site/*" might be unavoidable.
23:02:46  <str4d> And mapping urls:filenames is also unavoidable, because they are flat files, so the only way to get a url of a filename (without storing yet another map) is to use its path.
23:04:00  <topiltzin> zzz: which of the issues you brought up today would you say are real show-stoppers as opposed to incremental improvements?
23:04:03  <zzz> lang/site seems poor
23:04:07  <zzz> cant we just use content headers?
23:04:07  <str4d> General feeling I'm getting from reading some articles on SEO is that longer URLs with categorization are better.
23:04:48  <str4d> zzz, I recall discussing that with welterde, but I don't recall his argument against it.
23:06:05  <str4d> The backend just needs to get a language code from somewhere.
23:06:08  <str4d> @babel.localeselector
23:06:09  <str4d> def get_locale():
23:06:09  <str4d>     # If the language is already set from the url, use that
23:06:09  <str4d>     if hasattr(g, 'lang'):
23:06:09  <str4d>         return g.lang
23:06:09  <str4d>     # otherwise try to guess the language from the user accept
23:06:09  <str4d>     # header the browser transmits. The best match wins.
23:06:09  <str4d>     return request.accept_languages.best_match(['en', 'es', 'zh', 'de', 'fr', 'it', 'nl', 'ru', 'sv', 'cs', 'ar'])
23:06:09  <str4d> That's the current method.
23:06:24  <str4d> But IIRC welterde wasn't particularly happy with solely relying on ACCEPT_LANGUAGES
23:07:01  <zzz> i don't really know anything about the backend but it sounds like you're looking at a pretty rigid url->file mapping technology?
23:07:16  <zzz> i dont even know if you're using the same backend as now or something new
23:07:42  <str4d> zzz, it's not a rigid technology, no - it's the requirement of having the pages in mtn that is rigid.
23:07:46  <str4d> @app.route('/<string:lang>/site/<path:page>')
23:08:01  <str4d> ^-- that is the catcher for site urls.
23:08:31  <zzz> re: showstoppers, I wouldn't classify of my comments as showstoppers, only as comments. But it doesnt appear that we're at the point in the process yet to be talking about showstoppers anyway?
23:08:34  <str4d> The backend just turns "page" into a filesystem path and hands the resulting file to the template renderer.
23:09:01  <str4d> zzz, I'm hoping that discussion like this will help avoid showstoppers entirely =)
23:09:28  <zzz> right. but what sort of schedule are you on? days, weeks, months, years? I really don't know.
23:11:20  <str4d> I'd like to have the new site up within the next few months.
23:11:23  <str4d> And ideally the structure of the site will be sorted before the next release.
23:11:23  <str4d> well before*
23:11:23  <str4d> So the main showstopper I see is getting a cohesive design for the new structure.
23:12:01  <str4d> So weeks for the structure, months for the launch.
23:13:07  <str4d> Mainly because there are many other I2P-related projects I want to put time into, and I would like to finish some of my current ones first ^_^ but that said, I don't want to rush out a bad site.
23:13:50  <zzz> the structure of the site (menus, pages) is important, but not too hard. The structure of the files, as I said, is not the same thing, and not super-important, and I wouldn't spend more than about 5 more minutes on it.
23:16:16  <zzz> although, at the top level, getting rid of en/site would be great. seems like those shouldnt be in the urls
23:16:18  <str4d> zzz, I agree on that distinction, but under your current requirements the structure of the files is equivalent to the structure of the urls, which is very important for SEO.
23:16:48  <zzz> but with mapping you can skip over en/site for example.
23:18:24  <str4d> One thing - removing /en/site (and /en for the other pages) means that the new urls will get mixed up with the legacy catcher, which will make things tricky...
23:19:18  <zzz> it's important for seo but I dont know the right answers. Is en/site/docs/router/transport/udp/spec.html better or worse than /udp-specification.html? dunno.
23:19:25  <str4d> Which means that in addition to having a map of old legacy urls to new ones, we would need a map of new urls to their files, which is basically then a database.
23:20:12  <str4d> zzz, /docs/spec/ssu (or udp) is better than /udp-specification AFAICT from some preliminary SEO reading.
23:20:49  <str4d> Though now I'm reading a page advocating flat site architecture.
23:21:12  <str4d> ... or is it? I can't tell...
23:21:59  <zzz> there's ~575 files total, 180 meeting, 48 release, 95 status, leaving only 240 others
23:22:28  <str4d> "Along with smart internal linking, SEOs should make sure that the category hierarchy of the given website is reflected in URLs."
23:24:39  <str4d> The following is a good example of URL structure:
23:24:42  <str4d>     http://www.dmoz.org/Games/Video_Games/History/
23:24:42  <str4d> The following is a bad example of URL structure:
23:24:42  <str4d>     http://www.imdb.com/title/tt0468569/
23:24:45  <iRelay> Title: Open Directory - Games: Video Games: History (at www.dmoz.org)
23:24:53  <iRelay> Title: The Dark Knight (2008) - IMDb (at www.imdb.com)
23:26:52  <str4d> That seems to echo the general view I'm getting.
23:27:04  * str4d will do more SEO research... ugh...
23:27:07  <zzz> I think you gotta figure out the language thing first
23:27:34  <str4d> Yeah. Do we know that users will have ACCEPT_LANGUAGES set?
23:27:37  <str4d> s/that/if/
23:27:40  <iRelay> str4d meant: Yeah. Do we know if users will have ACCEPT_LANGUAGES set?
23:27:44  <str4d> Does the HTTP proxy filter that at all?
23:28:35  <zzz> if you handle accept-foo correctly then the crawlers will work better I would guess
23:28:35  <zzz> but what do you do about manual language selection - set a cookie and use that to override?
23:28:35  <zzz> i know almost nothing about how to do it
23:28:38  <zzz> until you decide for sure if it is lang/.... or .../foo_lang.html you can't get much further. I dont know if duck chose the en/site scheme but if he did it's not important that he did
23:29:25  <zzz> certainly for blogs, newspapers, etc the best practice is clear: myblog.com/2012/12/i-think-obama-is -the-whatever.html
23:32:14  <zzz> a lot of sites use short urls with redirects, for ads and for the url "guessers" - e.g. http://cbs.com/survivor
23:32:21  <iRelay> Title: Survivor: Watch Episodes and Video and Join the Ultimate Fan Community - CBS.com (at cbs.com)
23:32:40  <zzz> short urls for tweets and facebook and ads , maybe it doesnt matter if they redirect
23:33:39  <zzz> the i2p http proxy filters accept-*
23:45:41  <str4d>  zzz right, so there does need to be a URL-based lang option...
23:47:49  <str4d> cookies - is it okay to assume that all site users will have cookies enabled?
23:48:16  <str4d> or at least, all non-English users (both I2P and clearnet)
23:49:50  <zzz> dont know
23:54:23  <str4d> Certainly inside I2P, we don't want to alienate the more careful users.
23:58:45  <str4d> Anyway, this meeting has kinda turned into a website discussionfest - sorry!
23:58:45  <RN> right, I block cookies by default, and I think I'm not alone
23:59:07  <RN> it's important stuff str4d :)
00:02:24  <str4d> So I'll officially end the meeting now, but website discussion can continue (though I'm off for an hour or so).
00:02:27  <str4d> Unless anyone has any quick points they want to raise?
00:02:34  <str4d> 3
00:02:34  <str4d> 2
00:02:34  <str4d> 1
00:02:37  * str4d *baf*s the meeting closed.
00:07:57  <topiltzin> Thank everyone, this was good stuff.  I learned a lot about SEO and it's been joy to see zzz so enthusiastic. ( Could be the dg-pixie-dust aftereffects :) ) I wish I could contribute more but websites are not my thing at all.
00:08:13  <topiltzin> and thanks to KillYourTV and sighup-bot__ for the logs :)
00:13:54  <zzz> pfft. didnt know we were still in the meeting :)
00:26:40  <topiltzin> just zoned back in :)  back seat was so cozy I spaced out a little