[11:26:32] <Chiguireitor> guys, you tell me when i can fire the barrage of questions :D [11:26:50] <aa_> Chiguireitor: you make it sound so fun [11:27:04] <Chiguireitor> lol [11:29:49] <dunker> yes im ready ! [11:31:16] <+proppy_google> trying to get more people in :) [11:31:21] <Chiguireitor> hehehe [11:31:26] <dunker> wauw [11:31:33] <dunker> the google invasion [11:31:33] <jjg_google> brb - meeting [11:31:34] <Chiguireitor> using the lash and the chari proppy? [11:31:37] <dunker> haha [11:31:45] <Chiguireitor> chari -> chair [11:32:10] <aa_> wow, nice work proppy, I don't think I've ever seen this many googlers [11:32:20] <dunker> actually i dont have any questions [11:32:30] <Chiguireitor> i have 3 [11:32:33] <Chiguireitor> :D [11:32:38] <Chiguireitor> one is a stupid one btw [11:32:39] <dunker> k cool [11:32:45] <Chiguireitor> maybe you can solve it [11:32:47] <Chiguireitor> :D [11:32:56] <dunker> doubt ik ;] [11:33:00] <dunker> it [11:33:10] <Chiguireitor> heheheh [11:33:30] <Chiguireitor> well... i'm even asking something about free limits [11:33:34] <Chiguireitor> so you can join me hahahaha [11:33:34] <+proppy_google> wesc and gred said they will joined too [11:33:47] <Chiguireitor> ok... [11:34:03] <Chiguireitor> you just tell us when you're ready [11:34:20] <kevlar_google> are you going to /exec cat ~/ gaequestions.txt or something? lol [11:34:27] <dunker> haha [11:34:28] <kevlar_google> s/exec/oexec/ [11:34:31] <Chiguireitor> roflol [11:34:34] <dunker> would be cool ;] [11:34:36] <kevlar_google> or whatever the command is. [11:34:45] <Chiguireitor> could use dunker's bot [11:34:49] <Chiguireitor> for irc chats [11:34:52] <dunker> yeah [11:34:52] <Chiguireitor> lol [11:35:06] <kevlar_google> my bot idles on #go-nuts. [11:35:17] <dunker> ah a go bot ;] [11:35:23] <kevlar_google> of course! [11:35:29] <kevlar_google> IRC bots are the first things I write in a new language. [11:35:35] <dunker> yeah [11:35:38] <Chiguireitor> hahahaha [11:35:46] <dunker> did the same thing and im still programming it lol [11:35:53] <aa_> I always write a build system [11:36:06] <aa_> irc is too fiddly a protocol [11:36:06] <dunker> nice [11:36:18] <kevlar_google> they teach you string processing, networking, threading, file i/o, and of course all of the logic surrounding the interactions [11:36:21] <dunker> irc fiddly ? not really [11:36:40] <kevlar_google> yeah, especially from the client side, it's really straightforward [11:36:40] <aa_> it's not the most pleasant thing [11:36:42] <nullptr> aa_: itym FTP [11:36:46] <dunker> yeah you get strings thats it [11:37:30] <kevlar_google> you have a prefix, and you know it's there because of the :, you search for a " :" and if it's there you have a quoted argument, then you extract the rest by spaces and append the quoted argument and voila. [11:37:32] <dunker> but building bots to learn a area is a good thing .. [11:37:39] <dunker> did the same for GAE lol [11:38:02] <dunker> thats doesnt sound easy kevlar ;] [11:38:04] <nullptr> kevlar_google: do you guys run an internal irc server? [11:38:26] <dunker> ssssttt thats google internals [11:38:32] <dunker> not to be talked about ;] [11:38:35] <Chiguireitor> lol [11:38:36] <kevlar_google> dunker, it's like, 2 lines of python, or like 6 lines of go [11:38:54] <dunker> mine irc string to event converter is like 15 lines or so [11:39:00] <kevlar_google> (and like 650 characters / 10 lines of java) [11:39:05] <Chiguireitor> we run an openfire xmpp server here, and we have it talking to gtalk lol [11:39:06] <dunker> haha [11:39:08] <dunker> nurd ;] [11:39:12] <dunker> counting the chars [11:39:14] <dunker> lol [11:39:29] <dunker> my bot does xmpp as well ;] [11:39:30] <sahid> Hello appengine team, since 14hours i can not to send email from my app [11:39:33] <dunker> also gtalk [11:39:34] <kevlar_google> gtalk is nice, but I prefer IRC :) [11:39:37] <Chiguireitor> lol [11:39:42] <dunker> ah a real question [11:39:52] <Chiguireitor> can we fire up questions already!? [11:40:05] <dunker> if i want encrypted password protected chat then i use xmpp [11:40:07] <javamo> Hi guys, is ereporter tool working? It's not working for me [11:40:09] <kevlar_google> Chiguireitor, I haven't seen anyone from Google step up to organize this party [11:40:13] <dunker> lots of companies use xmpp internally [11:40:19] <Chiguireitor> :( [11:40:34] <+proppy_google> yes, we are still waiting for Greg or Wes to join [11:40:43] <dunker> so call them ! [11:40:45] <dunker> ;] [11:40:47] <Chiguireitor> dunker: yeah, more secure and better to use for internal communication on the go [11:40:48] <Chris_Schalk> I wouldn't wait for wes for greg [11:40:51] <dunker> waiting sucks ;] [11:40:58] <dunker> Chiguireitor: true [11:41:05] <dunker> i do xmpp on my mobile as well [11:41:09] <sahid> i'm pretty sur that come from appengine [11:41:18] <Chiguireitor> we here in the office have all androids setup with our internal xmpp server [11:41:20] <Chris_Schalk> sahid what is your appid? if you're ok to disclose [11:41:22] <h0x5f3759df> is it chat time with devs? [11:41:27] <nullptr> !8b are we being ignored? [11:41:27] <jsonbot> Better not tell you now. [11:41:34] <Chiguireitor> roflol [11:41:42] <sahid> Chris_Schalk, devel-inchallah [11:41:48] <Chris_Schalk> ok thanks [11:41:55] <dunker> oi sahid [11:41:55] <Chiguireitor> !8b are wes and greg coming? [11:41:56] <jsonbot> As I see it, yes. [11:42:02] <Chiguireitor> !8b you sure? [11:42:03] <jsonbot> You may rely on it. [11:42:04] <dunker> havent heared or seen you for a long time [11:42:11] <Chiguireitor> lol [11:42:11] <dunker> !chan-silent [11:42:12] <jsonbot> putting #appengine to silent mode [11:42:15] <sahid> hello dunker [11:42:23] <Chiguireitor> ahhh dunker.... :( [11:42:29] <Chiguireitor> you silenced our little bot [11:42:30] <dunker> hows going sahid still programming cool GAE stuff ? [11:42:33] <ozzyb> devel-in challah <--- tha t rocks lol! [11:43:18] <dunker> Chiguireitor: its not bot playing time now .. time for serious business ! [11:43:24] <Chiguireitor> hehehee :D [11:43:36] <dunker> otherwise just /msg the bot ;] [11:43:52] <Chiguireitor> btw dunker, have you seen the disparity of the channel quotas between the docs and the dashboard? [11:44:10] <+proppy_google> Chris_Schalk: let's go then :) [11:44:15] <sahid> Chris_Schalk, it's a real problem, i don't have update my app and i need email to validate member registraction [11:44:16] <dunker> ok question 1 for the google team: "when are we going to start office hour" [11:44:24] <dunker> ah proppy_google kicked it off [11:44:29] <Chiguireitor> cool [11:44:52] <Chiguireitor> are we going to just fire up questions or will it be federated by someone? say, proppy? [11:44:54] <Chris_Schalk> sahid - I see a lot of deadline exceeded exceptions for your inbox.. [11:45:05] <ozzyb> if __name__ == '__main__': start_already() [11:45:14] <dunker> so maybe a question ... is the working of the scheduler going to be open to see for everybody ? [11:45:20] <nvictor> hey all [11:45:24] <dunker> oi nvictor [11:45:40] <sahid> Chris_Schalk, what does that mean? [11:46:05] <nvictor> am i doing anything wrong? i subclassed a model, re-implemented put() updated a field and called super(). but it's not working [11:46:09] <Chiguireitor> googlers, there's a huge difference between the channel free quotas seen on the docs and the dashboard... what quota is the correct? [11:46:29] <Chiguireitor> dashboard says 46,310,400 [11:46:37] <Chiguireitor> docs 657,000 [11:46:49] <Chiguireitor> i don't have billing enabled (yet) [11:46:50] <marce_google> The dashboard is correct [11:46:53] <+proppy_google> Chiguireitor: I would say dashboard > doc [11:46:56] <marce_google> Please file a docs bug [11:46:59] <Chiguireitor> ok [11:47:07] <+proppy_google> Chiguireitor: feel free to fill a doc bug, with the doc is not correct [11:47:13] <+proppy_google> ah marce_google got me ;) [11:47:20] <Chiguireitor> hehehe [11:47:41] <Chris_Schalk> sahid - it means you are possibly taking longer than 30 seconds for this operation [11:47:59] <Chris_Schalk> has it worked before? [11:48:06] <Chiguireitor> ok, if i do a busy loop in a backend as fast as i can, and it hammers memcache a lot, will i get punished? say, with throttling or hit a memcach quota? [11:48:07] <sahid> yes [11:48:13] <sahid> Chris_Schalk, it's wierd... [11:48:18] <+proppy_google> dunker: I'm not sure what you mean by the "working of the scheduler", if you take a look at this year I/O presentation by guido and haugh, they do share detail about how the scheduler is working [11:48:35] <+proppy_google> let me found the link [11:48:55] <+proppy_google> dunker: http://www.google.com/events/io/2011/sessions/scaling-app-engine-applications.html [11:48:59] <dunker> proppy_google: the scheduler is going to be tuned to the new billing scheme .. as it start instances too much right now [11:49:19] <Chris_Schalk> sahid, yeah, something is taking too long when you view your email inbox [11:49:36] <dunker> since its so key to what bill ppl are going to get i was wondering if its algoritms are going to be set in stone ;] [11:49:36] <aa_> nvictor: that should work [11:49:49] <nvictor> hi aa [11:49:52] <sahid> Chris_Schalk, i send about 600 000emails per days [11:49:55] <aa_> hi nvictor :) [11:49:58] <nvictor> i am not sure what i'm doing wrong [11:50:10] <nvictor> i need to add that the model is an Expando though [11:50:20] <+proppy_google> dunker: I see no reason why they would share the scaling formula like they did in this talk, but you should ask Greg (coming in a few minutes) for confirmation [11:50:23] <dunker> guess we will have to wait till the thing is actually there [11:50:32] <sahid> Chris_Schalk, since 14hours none was sent [11:50:41] <dunker> yeah greg is the man i guess ;] [11:50:53] <marce_google> We are planning on giving people nobs to tune the scheduler [11:50:54] <kevlar_google> someone highlight me if there's a go question, I'm going to sift through mailing lists. :D [11:50:55] <aa_> nvictor: apart from that I never trust super to work, and I always explicitly call the superclass method, it should work [11:51:02] <marce_google> basically it's a trade of between new instances and latency [11:51:05] <+proppy_google> proppy_google: I also heard in this talk that they would potentially provide API to tune the formula [11:51:12] <dunker> haha proppy_google ;] [11:51:20] <nvictor> aa: ok will call the superclass [11:51:21] <aa_> haha, proppy is excited :) [11:51:21] <Chris_Schalk> sahid, the errors I see are when you view your inbox - maybe unrelated? [11:51:28] <+proppy_google> marce_google: you beat me again :) [11:51:29] <dunker> reminder to self [11:51:38] <Chris_Schalk> sahid, I would try a simple test [11:52:01] <dunker> marce_google: are there going to be admin knobs to tune that ? [11:52:08] <sahid> Chris_Schalk, yes [11:52:17] <sahid> Chris_Schalk, never mail are sent [11:52:18] <dunker> like i say i prefer high latency over extra instances [11:52:22] <dunker> ah well [11:52:26] <marce_google> yes. there will be admin nobs [11:52:28] <Chris_Schalk> sahid, you also want to make sure that you don't send too many at one time - http://code.google.com/appengine/docs/quotas.html#Mail [11:52:33] <dunker> ok cool [11:52:46] <Chiguireitor> does any connected googler cover backends? [11:52:47] <marce_google> the basic idea is the fewer instances, the more latency [11:52:53] <ozzyb> +1 for admin nobs [11:52:54] <dunker> yes [11:52:54] <marce_google> but some applications may be ok with that [11:53:12] <Chiguireitor> i would like to know if i do a busy loop in a backend as fast as i can, and it hammers memcache a lot, will i get punished? say, with throttling or hit a memcache quota? [11:53:18] <sahid> Chris_Schalk, yes i send about 500/minutes [11:53:21] <dunker> yeah guess the nature of the app determines what you need [11:53:36] <+proppy_google> how come the bot doesn't record +1 :) [11:53:42] <Chiguireitor> dunker turned it off [11:53:49] <Chiguireitor> bad dunker... bad bad dunker [11:53:50] <dunker> !chan-enable [11:53:54] <ozzyb> lol [11:53:55] <dunker> !chan-loud [11:53:56] <dunker> so [11:53:57] <Chris_Schalk> sahid, then I assume billing is enabled [11:53:57] <dunker> ;] [11:54:00] <jsonbot> done - chan-loud [11:54:03] <sahid> Chris_Schalk, last night i have sent 750 000emails [11:54:04] <dunker> well [11:54:10] <dunker> +1 doesnt work ++ does ;] [11:54:16] <dunker> google++ [11:54:17] <jsonbot> karma - google: 1 [11:54:21] <Chiguireitor> dunker++ [11:54:22] <jsonbot> karma - dunker: 1 [11:54:24] <Chiguireitor> yay [11:54:30] <sahid> Chris_Schalk, it's wierd, why 750 00 [11:54:45] <Chris_Schalk> sahid, so you successfully sent 750 last night? [11:54:46] <sahid> Chris_Schalk, Why just 750 00? [11:54:50] <sahid> Chris_Schalk, Why just 750000? [11:55:07] <Chris_Schalk> 750,000? [11:55:09] <sahid> Chris_Schalk, and no 750 003 [11:55:13] <sahid> Chris_Schalk, and no 756 003 [11:55:26] <sahid> Chris_Schalk, is not a problem with a quota? [11:55:59] <+proppy_google> Chiguireitor: I'm not sure about this [11:56:14] <+proppy_google> it looks like currently the memcache quota is unlimited [11:56:23] <+proppy_google> (an least that how it shows up in the dashboard) [11:56:29] <Chris_Schalk> sahid, it seems like it could be a quota problem [11:56:37] <Chiguireitor> proppy_google: yeah... but i will be using it for inter-instance/inter-backend communication [11:56:52] <+proppy_google> maybe greg would know more [11:56:57] <Chiguireitor> and i would prefer self-throttling than hitting a hard limit [11:57:01] <dunker> oi sudhirjonathan [11:57:08] <+proppy_google> Chiguireitor: beware that the key could vanish at any time [11:57:18] <Chiguireitor> yeah... it is used almost instantly [11:57:21] <Chris_Schalk> sahid, you're not sending spam are you? [11:57:23] <+proppy_google> so I would use http or protorpc for inter instance configuration [11:57:38] <+proppy_google> s/configuration/communication/ [11:57:43] <Chiguireitor> even i put a 100ms memcache validity for those keys [11:57:51] <Chiguireitor> don't want to pollute my cache [11:57:55] <sudhirjonathan> dunker, i'm here [11:58:00] <dunker> you could check if the key is lost and use http otherwise maybe [11:58:05] <sahid> Chris_Schalk, all are registered user but i don't know if my emails are bounced [11:58:13] <dunker> good thing sudhirjonathan ;] [11:58:26] <+proppy_google> yes, but still you could potentially lost a message and doesn't know anything about it [11:58:51] <sudhirjonathan> well, it's nice to feel wanted, but what is the good thing? [11:59:02] <Chiguireitor> well... i do an atomic increment of one memcache parameter... if i see a difference, i can tell if i lost messages [11:59:07] <Chris_Schalk> sahid, you could always include your own emails in your outbound messages [11:59:09] <dunker> dunno [11:59:14] <dunker> just saying hi [11:59:17] <+proppy_google> dunker: yep, but the other end doesn't necessarily know there was a message :) [11:59:18] <sahid> Chris_Schalk, last night has about 10pm (in france) all of my emails are not sent [11:59:29] <sudhirjonathan> hello :D [11:59:40] <dunker> well programming caches sucks [11:59:46] <dunker> ;] [12:00:02] <sahid> Chris_Schalk, i don't understand [12:00:11] <Chiguireitor> yeah... it would be better to have a proper message passing mechanism between isntances [12:00:14] <Chiguireitor> instances [12:00:17] <dunker> tried to implement instance memory cache once but you need invalidation scheme to make it work [12:00:23] <+proppy_google> Chiguireitor: oh so you are always using the same cache line for communicating ? [12:00:23] <dunker> which is hard todo [12:00:30] <dunker> Chiguireitor: yes [12:00:35] <sahid> Chris_Schalk, for example now i'm pay for emails not sent [12:00:35] <Chiguireitor> lemme pastebin that [12:01:07] <Chris_Schalk> sahid, do you know for sure that emails were not sent? [12:01:18] <sahid> Chris_Schalk: Recipients Emailed $0.0001/ Email 14% 280,850 of 2,002,000 [12:01:24] <sahid> Chris_Schalk, but none was send [12:01:38] <sudhirjonathan> dunker, I think i'm enough of an IRC noob for this to actually put a smile on my face. Greetings to you too, and to everyone here that I don't know but still feel a strong kinship with. [12:01:41] <+proppy_google> Chiguireitor: I think that's one of the goal of protorpc that will be available in 1.5.1 [12:01:47] <Chiguireitor> hehehe [12:01:52] <sahid> Chris_Schalk, yes ofcourse [12:01:58] <Chiguireitor> didn't know there was work in that area :D [12:01:59] <dunker> sudhirjonathan: ;] [12:02:04] <ozzyb> sudhirjonathan: cheers! [12:02:15] <dunker> ok i made up a new question [12:02:18] <+proppy_google> Chiguireitor: check today release not of the SDK pre-release ! [12:02:24] <+proppy_google> s/not/notes [12:02:30] <dunker> "is proppy_google going to be the new ikai?" [12:02:34] <Chris_Schalk> sahid, how are you sending your emails? via taks queues? [12:02:52] <+proppy_google> dunker: ikai is travelling right now, that's why he is not there [12:02:59] <dunker> ah yes ;] [12:03:03] <sahid> Chris_Schalk, yes via task queues [12:03:16] <+proppy_google> he is giving talks in Singapure :) [12:03:20] <dunker> proppy_google: almost thought you've taken over ;] [12:03:30] <Chiguireitor> where are the ProtoRPC docs? [12:03:30] <sudhirjonathan> proppy_google: is the protorpc version different from what in the trunk right now? [12:03:40] <dunker> spreading the word huh [12:03:59] <sudhirjonathan> Chiguireitor: https://docs.google.com/document/d/1QJBVWnbtYoiTFHIyqcBdnv5u4C1zYvDzFFGBIun35Ro/view?ndplr=1 [12:04:11] <+proppy_google> see previous blog post there: http://googleappengine.blogspot.com/2011/04/introducing-protorpc-for-writing-app.html [12:04:23] <sudhirjonathan> I <3 protorpc [12:04:29] <Chris_Schalk> sahid, my guess is that you are running into quota issues - perhaps exceeding 5100 recipients per minute [12:04:50] <+proppy_google> Chiguireitor: about messaging you could also rely on pull queue [12:05:08] <ozzyb> hmmm that protopc stuff looks interesting, must read... [12:05:14] <sudhirjonathan> proppy_google: The development server's datastore implementation now contains logic that [12:05:14] <sudhirjonathan> closely replicates the consistency guarantees of the High Replication [12:05:15] <sudhirjonathan> datastore - how does this work? [12:05:21] <Chris_Schalk> sahid, you need to make sure your not above: http://code.google.com/appengine/docs/quotas.html#Mail [12:05:35] <tkon> I'm trying to apply Brett Slatkin's advice for building scalable apps here http://www.google.com/events/io/2009/sessions/BuildingScalableComplexApps.html. In it he has a social network with stringlistpropertys for everybody's list of friends, and with it he can find mutual friends. Basically I'm wondering if I can still use this strategy if I use TWO stringlistpropertys comparing users likes and dislikes, and if it will still scale to p [12:05:59] <sudhirjonathan> ozzyb: Yeah, it's been hugely worth it for me... very conducive to building a client independent webapp [12:06:09] <sahid> Chris_Schalk, I don't see problem with my quota in dashboard [12:06:09] <+proppy_google> sudhirjonathan: feel free to ask your questions more openly as I might not be the only one able to answer :) [12:06:28] <sudhirjonathan> ah, okay... I don't see any other _googles around, though. [12:07:14] <Chiguireitor> http://pastebin.com/p65FzUCN [12:07:15] <Chris_Schalk> sahid, but you can guarantee that you're below the quota limits? [12:07:22] <Chiguireitor> protorpc seems similar [12:07:27] <Chiguireitor> but i'm already on GAE [12:07:37] <Chiguireitor> no need to communicate with the outside world [12:08:31] <sudhirjonathan> And this delay will be entirely random? [12:08:37] <Chiguireitor|2> sorry disconnected [12:09:07] <sudhirjonathan> random within limits, anyway.. [12:09:14] <marce_google> yes, random within limits [12:09:14] <sahid> Chris_Schalk, i'm pretty sure the problem comes from you [12:09:20] <gregd-google> Sorry I'm late to the party :) [12:09:26] <dunker> ;] [12:09:26] <Chiguireitor|2> hi greg [12:09:32] <Chris_Schalk> sahid, that sounds personal :-) [12:09:33] <sudhirjonathan> hello, greg [12:09:35] <dunker> so the real questions can start ... [12:09:39] <Chiguireitor|2> roflol [12:09:50] <kevlar_google> wb gregd-google [12:09:51] <gregd-google> oh, um, I think I'm in the wrong channel... [12:09:53] <gregd-google> ;) [12:09:58] <dunker> lol [12:10:00] <+proppy_google> hi gregd-google ! [12:10:03] <sahid> Chris_Schalk, yes... [12:10:08] <sudhirjonathan> is this the office hour or something? [12:10:13] <Chiguireitor|2> greg.... earlier i was asking a question [12:10:16] <Chiguireitor|2> sudhirjonathan: yep [12:10:17] <+proppy_google> now unleash the billing question flow! [12:10:21] <dunker> greg any idea on the ETA of the new billing ? [12:10:25] <sudhirjonathan> lol [12:10:25] <dunker> august ? [12:10:27] <dunker> september ? [12:10:31] <gregd-google> let the billing questions flow like wine! [12:10:34] <Chiguireitor|2> i would like to know if i do a busy loop in a backend as fast as i can, and it hammers memcache a lot, will i get punished? say, with throttling or hit a memcache quota? [12:10:37] <+proppy_google> sudhirjonathan: yes [12:10:43] <+proppy_google> 20 minutes remaining [12:11:06] <sudhirjonathan> the billing questions are actually tied in tight with threading, in my head [12:11:27] <dunker> yeah who i going to remove GIL from python lol [12:11:28] <sahid> Chris_Schalk, where is Guido! i need help ;) [12:11:32] <sudhirjonathan> can you guys confirm / deny that the new billing will come into effect only after / with threads? [12:11:50] <gregd-google> while we can't give an estimate when we will leave preview (which is when the new pricing kicks in), we are trying to get sample bills into 1.5.2 (1.5.1 is coming on next week) [12:12:07] <dunker> ah yes nice [12:12:08] <jrdesjardins> Google team, is it at all in your plans to provide a way to administer the URL's that are automatically created to point to previous versions an app [12:12:10] <Chiguireitor|2> yay for 1.5.1 [12:12:16] <sudhirjonathan> btw, awesome work on 1.5.1 [12:12:18] <Chiguireitor|2> google++ [12:12:18] <dunker> sample bllings sounds like a good idea [12:12:19] <jsonbot> karma - google: 2 [12:12:36] <dunker> man all the GAE programmers are so going to tune their apps to this [12:12:38] <jjg_google> google++ [12:12:39] <jsonbot> karma - google: 3 [12:12:48] <dunker> its like the early days all over again [12:12:49] <+proppy_google> jrdesjardins: not that I know of, did you fill a feature request ? [12:12:50] <gregd-google> sudhirjonathan Glad you liked it! [12:13:08] <jrdesjardins> nope, should I? [12:13:11] <sudhirjonathan> yeah - channel presence, protorpc, country headers = good stuff [12:13:21] <+proppy_google> sure, if enough people star it it could get some attention [12:13:22] <gregd-google> We should be exposing the correct datastore APIs in 1.5.1 as well so you can calculate what those will cost [12:13:29] <Chris_Schalk> sahid, just to be clear - you were successfully sending the same amount of emails before, well below quota limits, but then suddenly you can't send emails (not a single one)? and nothing else has changed? [12:13:34] <taaz> marce_google: is there a way to force the sdk eventual consistency behvaior to happen on demand for testing purposes? hard to test app behavior if it's really always random. would be nice to be able to shut it off too. [12:13:45] <Chiguireitor|2> gregd-google: can you tell me if memcache has a hard quota limit? i know it is "unlimited" in the dashboard, but i suppose there's a hard limit [12:13:47] <sudhirjonathan> how're you guys resolving countries without adding latency though? (or is that a huge secret?) [12:13:55] <gregd-google> I mean exposing the counts of them in the admin console, the APIs are not changing [12:14:15] <+proppy_google> jrdesjardins: one way to workaround this issue is to delete previous version, as johnlockwood said earlier [12:14:35] <+proppy_google> jrdesjardins: you could also overwrite application version to provide a catch all handler, that redirect to the correct version [12:14:43] <marce_google> you will be able to turn SDK eventually constincey on (and off) with a flag. But we want to replicate the actual behavior so right now you can't tune how random it will be [12:14:54] <sahid> Chris_Schalk, Yes, that is my problem since about 20hours now [12:15:03] <schuppe_google> taaz: the SDK will give you the chance to configure consistency behavior for testing [12:15:07] <jrdesjardins> proppy: ya, i know, it's not my ideal solution, but will have to do [12:15:20] <jjg_google> schuppe++ [12:15:21] <jsonbot> karma - schuppe: 1 [12:15:25] <sudhirjonathan> Reg. administering version urls, it would also help people (like me) writing different versions in python / go for different kinds of tasks [12:15:36] <Chris_Schalk> sahid, do you know that cause of your deadline exceeded exception? [12:15:53] <Chris_Schalk> sahid, (the) cause [12:15:57] <tkon> is this the wrong time to ask questions about the datastore? I'm not sure if I should repost my question. I was hoping a googler could help me [12:16:11] <Chiguireitor> :( also me :( [12:16:12] <dunker> tkon: this is the time ;] [12:16:21] <sudhirjonathan> tkon, if anyone knows the answer, they'll help [12:16:25] <jrdesjardins> btw, just cuz i've got the logs open in my browser right now, could we have log time relative to our local in the logger? [12:16:26] <tkon> dunker: lol [12:16:27] <gregd-google> Chiguireitor|2: I don't know the limit off the top of my head, I believe it is 100M [12:16:28] <+proppy_google> sudhirjonathan: you could already do that don't you ? [12:16:30] <Chris_Schalk> sahid, email system status appears ok [12:16:40] <+proppy_google> sudhirjonathan: one version in python, and one version in java [12:16:45] <sahid> Chris_Schalk, I don't have deadline in my log [12:17:05] <Chiguireitor> gregd-google: i'm asking because i'm doing this http://pastebin.com/p65FzUCN (the receiving end is a backend doing that non-stop) [12:17:05] <marce_google> jrdesjardins - check the 1.5.1 release notes :) [12:17:07] <sudhirjonathan> proppy_google: Yeah, I can, but I'm talking only from the POV of nicer urls, possibly public ones [12:17:15] <marce_google> (for the pre release SDK) [12:17:17] <+proppy_google> sudhirjonathan: version are strings [12:17:22] <sahid> Chris_Schalk, in my panel all are ok, and emails are sent [12:17:27] <gregd-google> dunker: I wish I could say precisely when it will be but we need to get a little closer to it to get more precise in our estimate [12:17:31] <+proppy_google> sudhirjonathan: so you could have myniceurl.myapp.appspot.com :) [12:17:33] <tkon> well googlers, here's my questions again ;) ... [12:17:34] <jrdesjardins> marce_google Great!, thx [12:17:35] <tkon> I'm trying to apply Brett Slatkin's advice for building scalable apps here http://www.google.com/events/io/2009/sessions/BuildingScalableComplexApps.html. In it he has a social network with stringlistpropertys for everybody's list of friends, and with it he can find mutual friends. Basically I'm wondering if I can still use this strategy if I use TWO stringlistpropertys comparing users likes and dislikes, and if it will still scale to p [12:17:51] <dunker> gregd-google: its just curiosity [12:18:00] <dunker> no need to know that right now ;] [12:18:07] <sudhirjonathan> proppy_google: Ah, okay, I'm on google apps, and want go-version.mysite.com and py-version.mysite.com [12:18:42] <kevlar_google> (woohoo, go!) [12:19:26] <sahid> Chris_Schalk, it's a chart of my users subscription [12:19:27] <gregd-google> Chiguireitor: Ah, so you are looking for total hard limit as well as how much data can be processed in a day? [12:19:29] <Chris_Schalk> sahid, I see deadline exceeded exceptions. I also see 'Exceeded soft process size' limit with.... [12:19:34] <johnlockwood> sudhirjonathan: cant you just make your version string "go-version" and this will make the url "go- version.appid.appspot.com [12:19:59] <sudhirjonathan> johnlockwood: will I be able to point to that directly from google apps, though? [12:20:28] <sahid> Chris_Schalk, yes that's comes with the my admins panel or if an user send a big imgae into the server... [12:20:36] <johnlockwood> sudhirjonathan: I don't know about that [12:20:44] <Chiguireitor> gregd-google: not so much data, because the amount of data is very low [12:21:07] <Chiguireitor> because i do an early test to not read much data if the counter hasn't been incremented [12:21:11] <Chiguireitor> but i read that counter a lot [12:21:15] <Chiguireitor> for each backend instance [12:21:20] <sudhirjonathan> johnlockwood: Yeah.. that's the prob. If I'm giving it out publicly, I'd like my-app-go-version.myservice.com [12:21:28] <Chiguireitor> btw, speaking of backend instances [12:21:43] <gregd-google> Chiguireitor: yea, that's what it looks like, I think the hard limit is just total data stored so you should be fine from that perspective, I don't know if we have a hard limit for the number of times you can read memcache in a day [12:21:49] <sudhirjonathan> tkon, could you give us an example of a query? [12:21:55] <+proppy_google> sudhirjonathan: yes, you right I just checked in the dashboard and it seems you could only map to the default application url [12:22:08] <+proppy_google> sudhirjonathan: so you would have to create different application for now [12:22:09] <sudhirjonathan> Yup :( [12:22:19] <Chiguireitor> can i have #_OF_BACKENDS x #_MAX_INSTANCES_PER_BACKEND instances running simultaneosly? [12:22:34] <sudhirjonathan> proppy_google: Can I share datastores, then :D ? [12:22:39] <Chiguireitor> gregd-google: cool [12:22:41] <Chris_Schalk> sahid, how big is your screenshot that you are sending via email? [12:22:46] <jjg_google> tkon: can you explain what you mean in a bit more detail? [12:22:57] <gregd-google> sudhirjonathan: this is a feature a few people have asked about, and frankly we think it would be useful as well [12:23:12] <+proppy_google> sudhirjonathan: hehe, another way to do it, it to do subdomain routing from the default version [12:23:16] <jrdesjardins> proppy_google: Do you think Control over version URL mapping is a good name for my feature request? [12:23:24] <sahid> Chris_Schalk, it's my users subscribes before the problem, i have remove the email validation http://i.imgur.com/dRaq5.png [12:23:36] <sudhirjonathan> Yeah... will star it, then... know if anyone has already created the feature request? [12:23:56] <+proppy_google> jrdesjardins: yes, sounds nice [12:24:12] <jrdesjardins> proppy_google: great thx [12:24:15] <sudhirjonathan> jrdesjardins: yes... and link here too? [12:24:18] <+proppy_google> jrdesjardins: don't worry if it doesn't get immediate attention, sometimes bug takes long to surface [12:24:23] <+proppy_google> since there is a lot of backlog [12:24:40] <sudhirjonathan> can campaign for it, though. [12:24:43] <gregd-google> Chiguireitor: you can have as many backends running simultaneously up to the current limits (documented at the bottom of this page: http://code.google.com/appengine/docs/python/backends/overview.html) [12:24:46] <+proppy_google> sudhirjonathan: you could have your application default version checking the hostname of the request [12:24:52] <Chris_Schalk> sahid, I don't understand your last statement [12:24:59] <+proppy_google> sudhirjonathan: and redirecting to the correct one [12:25:05] <Chiguireitor> gregd-google: that means i can have 100 instances running simultaneously? [12:25:28] <tkon> jjg_google: i'm not sure what more detail you want than this: http://pastebin.com/iXFkEJ3s.... what would you like to know? [12:25:42] <tkon> sudhirjonathan: i'll write up a query right now.. [12:25:47] <jjg_google> tkon: sorry - didn't see that link. looking now. [12:25:52] <sudhirjonathan> proppy_google: hmm. really don't want to go down that road... the point of having a separate go version was to have the really fast response times... want to use it for loading blocking javascript [12:26:08] <+proppy_google> sudhirjonathan: since you can alias more than one url to one app [12:26:10] <+proppy_google> sudhirjonathan: it would works [12:26:19] <gregd-google> Chiguireitor: yup! as long as they are chunked into 5 backends [12:26:26] <sahid> Chris_Schalk, sorry, To subscribe in my website, users must validate their email. [12:26:27] <Chiguireitor> cool [12:26:48] <Chiguireitor> gregd-google: its just to know if my implementation is "future-proof" [12:27:06] <+proppy_google> tkon: about the datastore question I don't have an answer on the top of my head [12:27:13] <Chris_Schalk> sahid, ok, is that why you are sending the image? [12:27:30] <jjg_google> tkon: I *could* be wrong, this is not my area of expertise, but I think that querying based on intersections of lists might not be do-able. [12:27:31] <+proppy_google> tkon: but I remember that there was some interesting design pattern, in this year io talk about the datastore [12:27:37] <+proppy_google> tkon: let me found you the link [12:27:38] <sudhirjonathan> proppy_google: That would add the latency of the python app (which is default) responding to and forwarding the request. The point of writing Go for me is superfast responses [12:27:43] <Chiguireitor> btw proppy_google, how's been the talks on issue 5123? [12:27:47] <sahid> Chris_Schalk, in this http://i.imgur.com/dRaq5.png you can see the problem, my users can't validate their email because they don't receive any email [12:27:54] <gregd-google> Chiguireitor: totally understandable, I'd expect over time those limits to go up, after all, that's the App Engine way :) [12:27:55] <sahid> Chris_Schalk, (sorry for my english) [12:28:00] <jjg_google> tkon: one of the notes in that talk was about how this only works when querying a single list, otherwise indexes "explode" [12:28:16] <+proppy_google> tkon: http://www.youtube.com/watch?v=xO015C3R6dw [12:28:20] <Chiguireitor> gregd-google++ [12:28:22] <sahid> Chris_Schalk, when i have see the problem, i have remove the email validation [12:28:28] <Chiguireitor> wes!! welcome to the party! [12:28:38] <+proppy_google> jjg_google: yes, I remember reading that on some stackoverflow questions too [12:28:38] <sahid> Chris_Schalk, but none emails was sent [12:28:49] <Wesley_Google> hello! i just arrived in the Zurich office :-) like... NOW [12:28:49] <Chris_Schalk> sahid, no worries :-) ok I this image is your own image of your email validation. I was actually asking if your were sending an email attachment to your users in bulk and how big that was. [12:29:20] <sudhirjonathan> _googles: speaking of issues getting noticed and starred, can we remove the issues that are definitely not going to be done from the tracker? [12:29:23] <Wesley_Google> Chiguireitor> did i reply back to you? i can't remember any more. backends in dev? [12:29:28] <+proppy_google> Chiguireitor: you could ask gregd-google I think he mentionned it to the team [12:29:29] <Chris_Schalk> sahid, so when you don't have email validation, the emails send ok? [12:29:31] <dunker> oi Wesley_Google [12:29:32] <Chiguireitor> proppy_google: yes [12:29:36] <Chiguireitor> Wesley_Google: yes [12:29:40] <+proppy_google> Wesley_Google: is there ! [12:29:47] <Wesley_Google> sudhirjonathan> removing them will mean people will open a new ticket :P [12:29:53] <sudhirjonathan> huzzah! Wesley_Google! [12:29:53] <Chiguireitor> lol [12:30:23] <Chiguireitor> Wesley_Google: i'm currently running my dev backends from the command line, they get killed properly there [12:30:24] <Wesley_Google> hi everyone, sorry i am late. i just traveling for the past 20 hours :P [12:30:38] <Wesley_Google> Chiguireitor> sounds good [12:30:57] <Chiguireitor> Wesley_Google: yeah... was thinking too windows-like [12:31:02] <+proppy_google> sudhirjonathan: there is a lot to do with the issue tracker in general [12:31:06] <sudhirjonathan> well, but python2.6 can go... we know 2.7 is coming. I don't think you guys will add C#. or PHP? [12:31:19] <+proppy_google> sudhirjonathan: I would like to get the community more involved but I'm not sure how [12:31:19] <gregd-google> sudhirjonathan: I know this is a crappy answer but you never know when something might eventually become a priority and the tracker helps us understand if we should work on it [12:31:21] <Chiguireitor> but it could be useful if that gets corrected [12:31:23] <ozzyb> sudhirjonathan: yeah, like the effing "PHP is a must" crap :p [12:31:25] <tkon> jjg_google: i was under the impression from Brett Slatkin's talk that comparing against a stringlistproperty of friends for mutual friend finding, worked because you'd have less than 5000 friends..... would my situation work if I limited my two stringlistproperty's to 70 each, as 70 * 70 = 4900, and therefore there couldn't be more than 5000 combinations of the two [12:31:29] <sahid> Chris_Schalk, Sorry but i don't understand, just none of my emails are sent since about 20hours i'm sur the problem comes from you because i don't have update a new version of my application [12:31:45] <gregd-google> sudhirjonathan: so we don't delete things that currently look like we wouldn't want to do them as things end up changing [12:31:52] <sahid> Chris_Schalk, and i see any error in my dashboard about that [12:31:58] <jjg_google> tkon: the issue is there is now way to filter for one attribute based on another attribute in the same query (to my knowledge) [12:31:59] <Wesley_Google> never say never... if there is enough demand but no promises! [12:32:08] <Chris_Schalk> sahid, and has the content of these outbound emails changed at all? or are they the same type of emails [12:32:17] <Chiguireitor> gregd-google sudhirjonathan: php on GAE gives me the creeps [12:32:17] <+proppy_google> gregd-google: maybe we could add a sometimemaybe label :) [12:32:28] <Chiguireitor> proppy_google++ [12:32:30] <jsonbot> karma - proppy_google: 1 [12:32:30] <+proppy_google> Chiguireitor: it is already possible with the java runtime @! [12:32:37] <sudhirjonathan> gregd-google: The flipside is that the issue tracker now becomes a feedback loop to itself [12:32:43] <Chiguireitor> proppy_google: yeah, i know.... the horror [12:32:50] <jjg_google> tkon: in the example query, the search is for a bunch of known things. Find people who have "me" and "otherguy" in their friends list, and also live in SF [12:32:57] <gregd-google> Chiguireitor: lol, we should do it just to get the communities talking to each other ;) [12:33:02] <sudhirjonathan> only the popular ones will be seen, unless people go looking specifically for stuff [12:33:15] <sahid> Chris_Schalk, nothing has changed [12:33:44] <jjg_google> tkon: in your query, the filter is based on the query itself (find me people who have the same items in their list). [12:33:46] <Chris_Schalk> all, anyone else have issues with outbound emails? [12:33:54] <sudhirjonathan> proppy_google: yeah, that can actually be marked as done, then. Just use the JVM. [12:34:00] <Chiguireitor> if php gets official support, i expect the latencies will get high because of the load [12:34:02] <jjg_google> tkon: the right-hand-side of the ='s is not known at query time [12:34:09] <jjg_google> tkon: does that make sense? [12:34:11] <Chiguireitor> and mostly for worthless forums with little activities [12:34:12] <Chiguireitor> lol [12:34:25] <gregd-google> sudhirjonathan: I know it might not always seem this way but we look at it often when we are deciding on priorities for what to work on next, although it is only one input, there are other things to consider, like not creeping out Chiguireitor [12:34:35] <sudhirjonathan> :D [12:34:39] <Chiguireitor> roflol [12:35:06] <tkon> jjg_google: what about a query like this.... http://pastebin.com/TGsUvMpD ... it's not perfect as i was hoping to find ANY matches of a like and dislike, but if i had to have the user lock down.... "i'm looking for anyone who likes dogs but doesn't like cats" that would be acceptable [12:35:06] <gregd-google> :) [12:35:09] <jjg_google> tkon: if you knew what to look for at query time, this would work... that is: you can search for "give me a person who dislikes dogs and likes cats". [12:35:14] <Chiguireitor> gregd-google: you know what would be nice.... sockets :D [12:35:16] <sahid> Wesley_Google, i have a big problem with my sent emails, you can investigate please? [12:35:21] <sudhirjonathan> hmm. does anyone even read all the 2592 requests? looks too much like an overload. [12:35:44] <+proppy_google> also feel free to report directly important bug #nnn to team member lurking here [12:35:50] <tkon> jjg_google: there yeah, that would be acceptable, sort of [12:35:52] <jjg_google> tkon: Exactly -- that seems fine. You'd have to ask someone more knowledgeable about the indexing numbers [12:36:05] <jjg_google> tkon: but that looks like a perfectly valid query to me [12:36:14] <+proppy_google> sound like a bit unfair and mafia-like, but we can put it on the top of the pile to report to engs [12:36:14] <tkon> jjg_google: where might i go to find such a person? [12:36:35] <Chris_Schalk> sahid, we'll have a chat about your emails [12:36:56] <Chiguireitor> sudhirjonathan: proppy reads them..... eventually [12:36:57] <Chiguireitor> lol [12:37:05] <+proppy_google> sudhirjonathan: I'm tried to triage a few of those each day [12:37:06] <jjg_google> tkon: To be honest I'd post the question on Stack Overflow -- nickjohnson seems to be the most knowledgeable person I've met about all things AppEngine [12:37:07] <gregd-google> Chiguireitor: yea, that would be wouldn't it ;) [12:37:14] <+proppy_google> but the back log is enourmous [12:37:14] <Wesley_Google> yes sahid... what chris said :-) [12:37:28] <gregd-google> sudhirjonathan: that's why starring actually does help [12:37:33] <sudhirjonathan> tkon: Yeah, SO sounds like a good idea.. .link here too [12:37:37] <+proppy_google> 730 New issues (not triaged) [12:37:37] <+proppy_google> 1729 Acknowledges (read) [12:37:37] <+proppy_google> 95 Accepted (internal bug filled) [12:37:40] <+proppy_google> FYI [12:37:54] <Chris_Schalk> sahid, we have your app id.. so we'll look into it further [12:38:00] <Chiguireitor> proppy_google++ for being a nice triager [12:38:01] <jsonbot> karma - proppy_google: 2 [12:38:05] <+proppy_google> tkon: yes, there is already a few question about datastore best practice on SO [12:38:09] <sudhirjonathan> nice... will start starring, then [12:38:17] <jrdesjardins> For thos that might be interested. Here is a link towards the feature request I filed concerning "Control over version URL mapping" http://code.google.com/p/googleappengine/issues/detail?id=5202 [12:38:28] <+proppy_google> I will graph those each weak [12:38:31] <+proppy_google> and see if there is progress [12:38:46] <gregd-google> sudhirjonathan: we try to read them all, then once we are done, someone just creates a new one! [12:38:49] <sahid> Chris_Schalk, Wesley_Google ok thank a lot [12:38:54] <dunker> haha [12:39:02] <+proppy_google> anyways nice chat all! [12:39:03] <jrdesjardins> sry :-) [12:39:26] <dunker> yeah we should do it more often !! [12:39:29] <+proppy_google> if some of you are coming to europython in 2 weeks, there will be some of us there [12:39:32] <dunker> haha [12:39:33] <sudhirjonathan> gregd-google: Yeah, I'm surprised a fairly small team can (or is trying) handle all the GAE developers in the world. [12:39:38] <dunker> so everyday a 6 CET ... [12:39:41] <+proppy_google> Wesley_Google: aa_ : and me [12:39:45] <tkon> jjg_google: also, i just want to be clear...i'm not looking for find matches with my WHOLE lists with another person... it's more like...if I like dogs pigs and squirrels, and dislike cats, elephants and cows, find me the people that like ONE of the things I don't, and dislikes ONE of the things I like [12:39:59] <Chiguireitor> dunker: no dunker, you cannot ask about billing everyday at 6 CET [12:40:02] <Chiguireitor> lol [12:40:05] <tkon> jjg_google: is that not possible either you dont' think? [12:40:07] <+proppy_google> dunker: I will be lurking there, so feel free to ping me [12:40:10] <dunker> hehe Chiguireitor ;] [12:40:11] <colarte> hello, are there any #appengine irc logs posted anywhere?? [12:40:14] <Chris_Schalk> sahid, no problem, we're already chatting about it.. [12:40:15] <jjg_google> tkon: [12:40:15] <gregd-google> sudhirjonathan: :) we try our best! [12:40:17] <dunker> proppy_google: cool [12:40:24] <gregd-google> I gotta run, good talking to you all [12:40:27] <dunker> about time google got a liason in here ;] [12:40:31] <Chiguireitor> bye! [12:40:33] <dunker> ltrs gregd-google [12:40:34] <+proppy_google> *_google: thanks a lot for being there [12:40:35] <sudhirjonathan> jrdesjardins: starred [12:40:37] <jjg_google> tkon: the issue is that you need to pull the things you like out first and then issue a separate query to find those other people [12:41:05] <sudhirjonathan> thanks a got, _googles. And great work on 1.5.1. Looking forward. [12:41:18] <jjg_google> tkon: so if you first say "here's the list of things that I like and dont like" [12:41:40] <+proppy_google> argggg, how to copy thing in ircloud [12:42:05] <+proppy_google> can someone email the transcript to the groups ? [12:42:10] <jjg_google> tkon: and then issue a query, "find people that like=thing1, like=thing2, dislike=thing1" -- that should be the same as where like = thing1 or like = thing2 and dislike = thing3 [12:42:32] <+proppy_google> irccloud-- [12:42:33] <jsonbot> karma - irccloud: -1 [12:42:40] <ozzyb> haha [12:42:48] <Chiguireitor> tkon: why don't you use a publisher- subscriber pattern? [12:42:51] <ozzyb> btw, thanks again googlers! [12:43:17] <+proppy_google> Chiguireitor: got the full transcript ? [12:43:48] <Chiguireitor> want me to do that? [12:43:59] <Chiguireitor> or do i pm you? [12:44:08] <+proppy_google> feel free to post it directly to the groups [12:44:25] <dunker> clever ;] [12:44:25] <Chiguireitor> :) [12:44:33] <jjg_google> proppy++ thanks for organizing [12:44:33] <jsonbot> karma - proppy: 1 [12:44:40] <jjg_google> bye guys! [12:44:44] <Chiguireitor> bye [12:44:45] <dunker> ltrs jjg_google [12:44:52] <Wesley_Google> thx everyone for coming! [12:44:53] <+proppy_google> jjg_google++ for attending [12:44:54] <jsonbot> karma - jjg_google: 1 [12:44:54] <Chiguireitor> yeah the party had some nice drinks
-- You received this message because you are subscribed to the Google Groups "Google App Engine" group. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.
