| Time |
S |
Nick |
Message |
| 00:00 |
|
tjsnell |
what did you try? |
| 00:02 |
|
|
AlCoF joined ##javaee |
| 00:02 |
|
cem_ |
tjsnell: there are different software for this but none worked for hand written to digital |
| 00:03 |
|
tjsnell |
because that's a very hard issue |
| 00:03 |
|
cem_ |
sfisque: how can i do it ? |
| 00:04 |
|
sfisque |
assuming you have a scanner, write a JNI wrapper around a twain driver (twain is a standard scanner API that most (all?) scanner manufacturers abide by when writing their drivers) |
| 00:04 |
|
tjsnell |
find some software that can handle hand writing |
| 00:05 |
|
tjsnell |
his issue is ocr |
| 00:05 |
|
sfisque |
oh, are you wanting OCR |
| 00:05 |
|
sfisque |
maybe google around and see if someone has open source OCR library (doubtful, but plausible) |
| 00:05 |
|
cem_ |
Yes tjsnell right on the point |
| 00:05 |
|
sfisque |
if so, if its java, there you go. if it's c/c++, you can still go the JNI route |
| 00:06 |
|
tjsnell |
especially for hand writing |
| 00:06 |
|
tjsnell |
jni is the path of much woe |
| 00:06 |
|
sfisque |
it's not that bad, but not for the feint of heart |
| 00:06 |
|
sfisque |
thar be dragons there.... |
| 00:07 |
|
tjsnell |
oh it's easy at first |
| 00:08 |
|
sfisque |
reminds me of taht scene with jeff goldblum in jurassic park 2. "oh, you say that now… but later… there's lots of running, and screaming...." |
| 00:09 |
|
pdurbin |
maybe opencv? https://github.com/Itseez/opencv/blob/master/samples/cpp/letter_recog.cpp |
| 00:14 |
|
cem_ |
thanks pdurbin |
| 00:17 |
|
cem_ |
i'll inform you guys if things go fine :) |
| 00:17 |
|
cem_ |
bye |
| 00:22 |
|
tjsnell |
if we don't hear from him I guess it's a fail |
| 00:23 |
|
pdurbin |
it's a very hard issue |
| 00:43 |
|
|
AlCoF joined ##javaee |
| 00:57 |
|
|
sfisque joined ##javaee |
| 00:58 |
|
|
sfisque1 joined ##javaee |
| 00:59 |
|
|
nick___ joined ##javaee |
| 01:01 |
|
nick___ |
hello |
| 01:02 |
|
nick___ |
what are different types of webservice are there ? |
| 01:03 |
|
nick___ |
JAX RPC , JAX WS , SOAR , JMS, Apache AXIS....................... ?? |
| 01:06 |
|
tjsnell |
about a bazillian |
| 01:06 |
|
sfisque1 |
JMS is not a web service. axis is just an impl of SOAP |
| 01:06 |
|
tjsnell |
Apache Axis isn't a webservice |
| 01:06 |
|
sfisque1 |
basically jax-rpc, jax-ws, soap, and rest are your basic choices |
| 01:07 |
|
sfisque1 |
as far as "standards" are concerned |
| 01:07 |
|
tjsnell |
websockets |
| 01:07 |
|
sfisque1 |
after that you have a ton of "proprietary" solutions, hessian, http invocation (spring). |
| 01:07 |
|
sfisque1 |
is websockets really considered web services? |
| 01:07 |
|
tjsnell |
not sure |
| 01:08 |
|
sfisque1 |
javaeebot lucky websockets specification |
| 01:08 |
|
sfisque1 |
oh noze |
| 01:08 |
|
sfisque1 |
javaeebot` lucky websockets specification |
| 01:08 |
|
javaeebot` |
sfisque1: http://tools.ietf.org/html/rfc6455 |
| 01:09 |
|
sfisque1 |
what i thought. it's just two-way comm over http. transport protocol not a service layer |
| 01:09 |
|
sfisque1 |
quote : The WebSocket Protocol enables two-way communication between a client |
| 01:09 |
|
sfisque1 |
running untrusted code in a controlled environment to a remote host |
| 01:09 |
|
sfisque1 |
that has opted-in to communications from that code. The security |
| 01:09 |
|
sfisque1 |
model used for this is the origin-based security model commonly used |
| 01:09 |
|
sfisque1 |
by web browsers. The protocol consists of an opening handshake |
| 01:09 |
|
sfisque1 |
followed by basic message framing, layered over TCP. The goal of |
| 01:09 |
|
sfisque1 |
this technology is to provide a mechanism for browser-based |
| 01:09 |
|
sfisque1 |
applications that need two-way communication with servers that does |
| 01:09 |
|
sfisque1 |
not rely on opening multiple HTTP connections (e.g., using |
| 01:09 |
|
sfisque1 |
XMLHttpRequest or <iframe>s and long polling). |
| 01:10 |
|
sfisque1 |
so you "can" do web services over websockets, but it is just a transport mech |
| 01:13 |
|
|
AlCoF___ joined ##javaee |
| 01:22 |
|
nick___ |
amazon ws ? |
| 01:22 |
|
nick___ |
where should i start |
| 01:23 |
|
nick___ |
to learn web service |
| 01:23 |
|
pdurbin |
aws is a thing |
| 01:23 |
|
pdurbin |
nick___: what are your goals? |
| 01:23 |
|
nick___ |
pdurbin: what the heck are you asking ? |
| 01:23 |
|
pdurbin |
nick___: what are you trying to learn? |
| 01:24 |
|
nick___ |
i want know about web service pdurbin |
| 01:28 |
|
pdurbin |
when Twitter was new you could just curl their API. now you have to auth |
| 01:29 |
|
pdurbin |
I guess GitHub still lets you play with parts of their API without a password |
| 01:29 |
|
pdurbin |
nick___: I'd start there |
| 01:33 |
|
pdurbin |
play with the data at a url such as https://api.github.com/users/arun-gupta/repos |
| 01:34 |
|
pdurbin |
docs at http://developer.github.com/v3/repos/#list-user-repositories |
| 01:38 |
|
tjsnell |
heh |
| 01:38 |
|
tjsnell |
I'm writing a camel github example now |
| 01:42 |
|
pdurbin |
tjsnell: how to use camel? |
| 01:43 |
|
tjsnell |
just throwing up short camel examples at code.notsoclever.cc |
| 01:43 |
|
sfisque1 |
what are you automating? github fetches? commit reports? |
| 01:43 |
|
sfisque1 |
oh |
| 01:43 |
|
tjsnell |
oh, in this case I'm putting a SOAP front end on github REST |
| 01:43 |
|
tjsnell |
just demoing a bridge of the 2 |
| 01:44 |
|
tjsnell |
the github api was just a convenient example to use |
| 01:44 |
|
sfisque1 |
so, you're consuming REST and rebundling it? or is this a XSLT massage? |
| 01:45 |
|
nick___ |
my team work uses axis so i want to know about webservice |
| 01:45 |
|
tjsnell |
exposing git API as a SOAP interface |
| 01:45 |
|
tjsnell |
currently the payload is still a json string embedded in the xml package :) |
| 01:45 |
|
tjsnell |
that's up next |
| 01:45 |
|
sfisque1 |
ah, so just a wrapper |
| 01:46 |
|
tjsnell |
yeah |
| 01:46 |
|
tjsnell |
currently we have an app internally that only communicates over SOAP |
| 01:46 |
|
tjsnell |
so just showing protocol bridging |
| 01:46 |
|
sfisque1 |
aye |
| 01:49 |
|
pdurbin |
javaeebot: lucky apache axis |
| 01:49 |
|
javaeebot |
pdurbin: http://axis.apache.org/axis/ |
| 01:50 |
|
sfisque |
javaeebot lucky apache axis first tutorial tomcat |
| 01:50 |
|
javaeebot |
sfisque: http://axis.apache.org/axis/java/install.html |
| 01:51 |
|
nick___ |
i'm confused i have no clue on web service wsdl,rpc etc i dont get it |
| 01:51 |
|
tjsnell |
pick a tutorial and learn that |
| 01:51 |
|
tjsnell |
then another |
| 01:51 |
|
tjsnell |
REST certainly a bit easier |
| 01:51 |
|
tjsnell |
especially with Jackson |
| 01:51 |
|
tjsnell |
and Jersey |
| 01:52 |
|
tjsnell |
or Play |
| 01:52 |
|
tjsnell |
or Camel |
| 01:52 |
|
tjsnell |
... |
| 01:52 |
|
pdurbin |
I still say... go play with a REST API like GitHub's... get a feel for it |
| 01:53 |
|
pdurbin |
doesn't java ee let me do a lot of what axis is trying to do? help me build APIs? |
| 01:54 |
|
tjsnell |
axis doesn't do rest does it? |
| 01:54 |
|
tjsnell |
CXF does |
| 01:54 |
|
tjsnell |
which kills axis anywya |
| 01:55 |
|
tjsnell |
Jersey is the jee rest impl |
| 01:55 |
|
tjsnell |
jackson the json impl |
| 01:56 |
|
pdurbin |
yeah, seems like axis does soap |
| 01:57 |
|
sfisque |
i better. that's what it's for |
| 01:57 |
|
sfisque |
:P |
| 01:57 |
|
tjsnell |
cxf does soap better |
| 01:57 |
|
sfisque |
@WebService. easy. done. |
| 01:58 |
|
pdurbin |
sounds like nick___ should learn soap if it's what his team is using |
| 01:59 |
|
nick___ |
pdurbin: they use both each for different project 1 new and other is old |
| 02:01 |
|
pdurbin |
nick___: both? soap and rest? |
| 02:01 |
|
nick___ |
yes |
| 02:02 |
|
nick___ |
1 started on 2007 they use soap something like that |
| 02:02 |
|
tjsnell |
sfisque: https://gist.github.com/tjsnell/e885dfac7eab108569b1 |
| 02:02 |
|
tjsnell |
from my camel project :) |
| 02:03 |
|
pdurbin |
nick___: ok. makes sense |
| 02:05 |
|
sfisque |
nifty |
| 04:01 |
|
|
nick___ left ##javaee |
| 05:28 |
|
|
bobek joined ##javaee |
| 05:29 |
|
|
jenue joined ##javaee |
| 05:44 |
|
|
dangertools joined ##javaee |
| 05:44 |
|
|
dangertools joined ##javaee |
| 06:42 |
|
|
Quest joined ##javaee |
| 06:46 |
|
Quest |
sess hi |
| 06:47 |
|
Quest |
whartung sfisque , we agreed yesterday that we cant send / recieved mere byte[]s via http. right? there has to be either string or multipart. and thats it.? |
| 06:54 |
|
|
cheater_3 joined ##javaee |
| 07:04 |
|
Quest |
sess whartung sfisque line 947 http://pastebin.ca/2472645 |
| 07:11 |
|
sess |
Quest: im not very familiar with http connections from java, so dont think i can help much regardless of the issue |
| 07:11 |
|
sess |
also off to work |
| 07:22 |
|
Quest |
ok. thx |
| 08:27 |
|
|
liuzhoou joined ##javaee |
| 09:00 |
|
|
Radditz joined ##javaee |
| 09:39 |
|
|
Radditz joined ##javaee |
| 10:09 |
|
pdurbin |
Quest: application/octet-stream consists of "mere bytes" |
| 10:09 |
|
Quest |
ya |
| 10:09 |
|
Quest |
a byte [] inside the http.responst.input/outputStream() would be a valid http ? I was in an image that only text and multipart can be uploaded and the downloads are text/html and image/data with headers |
| 10:15 |
|
|
Radditz joined ##javaee |
| 10:20 |
|
Quest |
Radditz welcome back |
| 10:20 |
|
Radditz |
thank you Quest , my wifi connection drops often |
| 10:20 |
|
Quest |
so does mine |
| 10:24 |
|
|
bobek left ##javaee |
| 10:56 |
|
|
Quest_ joined ##javaee |
| 10:59 |
|
|
Quest joined ##javaee |
| 11:09 |
|
|
MegaMatt joined ##javaee |
| 11:54 |
|
|
kotten joined ##javaee |
| 12:26 |
|
|
hurk joined ##javaee |
| 12:48 |
|
|
tommmied joined ##javaee |
| 13:24 |
|
pdurbin |
whartung: dunno if you're awake yet but when I add your exclusions I get the same error about Provider org.apache.xerces.jaxp.DocumentBuilderFactoryImpl not found |
| 13:24 |
|
pdurbin |
Exception while loading the app : java.lang.IllegalStateException: ContainerBase.addChild: start: org.apache.catalina.LifecycleException: java.lang.RuntimeException: javax.xml.parsers.FactoryConfigurationError: Provider org.apache.xerces.jaxp.DocumentBuilderFactoryImpl not found |
| 13:24 |
|
zoot |
oh |
| 13:24 |
|
zoot |
check the systemproperty |
| 13:25 |
|
zoot |
and set it to something else |
| 13:25 |
|
zoot |
or add that jar |
| 13:25 |
|
zoot |
:P |
| 13:26 |
|
pdurbin |
zoot: when I add that jar I get "DTD factory class org.apache.xerces.impl.dv.dtd.DTDDVFactoryImpl does not extend from DTDDVFactory" as I explained at http://irclog.greptilian.com/javaee/2013-10-29#i_35909 |
| 13:26 |
|
pdurbin |
I'm sure people here are sick of hearing about it. :) |
| 13:26 |
|
pdurbin |
but I wanted to let whartung know I tried his thing |
| 13:27 |
|
zoot |
welcome to jarhell |
| 13:27 |
|
pdurbin |
hot enough for you? |
| 13:27 |
|
zoot |
i'm fraid we can't find your class to day |
| 13:35 |
|
pdurbin |
it's like school during a fire drill. no class! -- that kid on Fat Albert |
| 13:35 |
|
|
Naros joined ##javaee |
| 13:40 |
|
|
aldaris joined ##javaee |
| 13:40 |
|
|
aldaris left ##javaee |
| 13:41 |
|
|
balo joined ##javaee |
| 13:41 |
|
balo |
hi |
| 13:41 |
|
Naros |
hello |
| 13:42 |
|
|
tmichel joined ##javaee |
| 13:43 |
|
pdurbin |
balo: sorry about the confusion over #javaee vs. ##javaee: http://irclog.perlgeek.de/openam/2013-10-30#i_7788261 |
| 13:43 |
|
balo |
pdurbin: np, thanks for the tip. i didn't know about this channel :) |
| 13:44 |
|
pdurbin |
I need to write a patch for philbot: http://irclog.perlgeek.de/ilbot/2013-10-20#i_7739131 |
| 13:45 |
|
balo |
:) |
| 14:10 |
|
zoot |
http://devnull-as-a-service.com/ |
| 14:12 |
|
balo |
i saw this on twitter today :D i'll definetly use it... the devnull on my machine isn't scalable :( |
| 14:12 |
|
balo |
i hope they use nodejs |
| 14:17 |
|
zoot |
but you can use a clouddatabase like a filesystem with nodejs!!!1 |
| 14:30 |
|
|
AlexCzar joined ##javaee |
| 14:43 |
|
|
polosatinsk joined ##javaee |
| 14:55 |
|
|
kobain joined ##javaee |
| 15:03 |
|
|
kotten joined ##javaee |
| 15:14 |
|
|
AlexCzar joined ##javaee |
| 15:44 |
|
|
jieryn joined ##javaee |
| 15:44 |
|
|
jieryn joined ##javaee |
| 16:27 |
|
Naros |
Anyone here use JPA2.0 criteria? |
| 16:33 |
|
sfisque |
yes |
| 16:33 |
|
sfisque |
we only use it |
| 16:33 |
|
sfisque |
no jpql |
| 16:33 |
|
Naros |
Cool, got a conundrum. |
| 16:33 |
|
sfisque |
kk |
| 16:34 |
|
Naros |
EntityA has an EntityB which has an EntityC |
| 16:34 |
|
Naros |
I am querying EntityA but also need to eagerly get B & C |
| 16:34 |
|
Naros |
I've figured out how to get B, but no idae how to get C |
| 16:35 |
|
Naros |
Join<WorkOrder, Facility> workOrderFacility = workOrderRoot.join(WorkOrder_.plant); |
| 16:35 |
|
Naros |
but i need Site inside Plant too |
| 16:35 |
|
Naros |
er inside Facility |
| 16:35 |
|
sess |
select a from A join fetch a.b b join fetch b.c c |
| 16:35 |
|
Naros |
yah but that's using JPQL |
| 16:35 |
|
sess |
joins does not fetch |
| 16:35 |
|
sess |
fetches fetch |
| 16:35 |
|
Naros |
I'm using the JPA2.0 CriteriaBuilder API |
| 16:36 |
|
sess |
use fetch instead of join |
| 16:36 |
|
Naros |
oh |
| 16:36 |
|
Naros |
i gotcha |
| 16:36 |
|
sfisque |
depends. do you need to specificy qualifiers on the attributes of B and C. because you cannot add criteira on the fetches |
| 16:36 |
|
sfisque |
do you need to liimit on B.x or C.y |
| 16:36 |
|
sess |
if you need that, you both join and join fetch |
| 16:36 |
|
sess |
in any case, he needs to fetch |
| 16:36 |
|
sfisque |
aye, but be careful with that, using both in tandem can create "pathalogical" queries |
| 16:37 |
|
Naros |
In this case no, just need to load the entitys for the view is all |
| 16:37 |
|
sfisque |
it IRKS me that fetchs do not have the ability to limit on atrributes |
| 16:37 |
|
sfisque |
ok, then have at it |
| 16:37 |
|
sfisque |
:-) |
| 16:37 |
|
sess |
sfisque: pathalogical? |
| 16:38 |
|
sfisque |
we've had a few queries that jpa constructed double and triple redundant joins which confused oracle's explain planner and basically returned in horrible time |
| 16:38 |
|
sfisque |
we had to rewrite the queries to slice the data differently |
| 16:38 |
|
sess |
it might double join when using both join and join fetch |
| 16:38 |
|
sess |
buuut you wont get away from that i guess |
| 16:39 |
|
sfisque |
aye. adn depending on how the whole thing is assembled into sql, it can confuse the rdmbs explain planner |
| 16:39 |
|
Naros |
sfisque: do you guys used the type-safe api or just strings? |
| 16:39 |
|
sfisque |
***rdbms |
| 16:39 |
|
sess |
how would it, joins are simple stuff |
| 16:39 |
|
sess |
its just normal PK-FK joins |
| 16:39 |
|
sfisque |
typesafe metadata |
| 16:39 |
|
sess |
Naros: i use typesafe |
| 16:39 |
|
sess |
no reason to use criteria api without it |
| 16:39 |
|
Naros |
hm, how would I construct the type-safe fetch statement for C |
| 16:39 |
|
sess |
you add the metamodel generator plugin |
| 16:40 |
|
sess |
Entity_.field |
| 16:40 |
|
Naros |
workOrderRoot.fetch(WorkOrder_facility); works for the child. |
| 16:40 |
|
sfisque |
for C? |
| 16:40 |
|
Naros |
but for site, I dont see anything |
| 16:40 |
|
sfisque |
oh class C |
| 16:40 |
|
sess |
workOrderRoot.fetch(WorkOrder_facility).fetch(Facility_.thatOtherCollection); |
| 16:40 |
|
Naros |
k |
| 16:40 |
|
sess |
you can chain them just like regular joins |
| 16:41 |
|
sfisque |
aye |
| 16:41 |
|
sess |
make sure to remember JoinType.Left if needed |
| 16:41 |
|
Naros |
ty, sorry the API is new to me since we haven't used it before and the Hibernate Criteria API is deprecated so making the move to it ourselves. |
| 16:41 |
|
sess |
its quite complicated to learn |
| 16:41 |
|
Naros |
yah im finding out :( |
| 16:41 |
|
sess |
i still have to look up things after several years usage |
| 16:42 |
|
sess |
using group/having/subqueries is witchcraft |
| 16:42 |
|
sfisque |
jpa criteria are "almost" 1-1 with hibernate. hibernate team contributed heavily to jpa 2.0 jsr |
| 16:42 |
|
Naros |
aye, there is a lot of overlap for sure. |
| 16:42 |
|
Naros |
It's more the syntax |
| 16:42 |
|
sfisque |
yup |
| 16:42 |
|
Naros |
We heavily used Criteria API from Hibernate in the prior release of our app |
| 16:42 |
|
sfisque |
and LCD. hibernate is denser in features, jpa is more general |
| 16:43 |
|
Naros |
It's pretty powerful but the issue is there were many bugs when it came to things like nationalization :E |
| 16:43 |
|
Naros |
One I found recently where a nationalized column could not be equality checked with using case insensitiity :( |
| 16:45 |
|
Naros |
I don't particularly care for the necessity of adding '%' when I want to do begins with/ends with/contains type operations tho |
| 16:48 |
|
Naros |
Does your metamodel generator create a class for your @EmbeddedId's ? |
| 16:49 |
|
Naros |
e.g. want to query on an entity composite key specific property. |
| 16:50 |
|
Naros |
only way I've found so far is root.get("pk").get("propertyName") |
| 16:51 |
|
sess |
arent embedded ids pretty transparent |
| 16:51 |
|
sess |
i.e would not the fields be included in the parent |
| 16:52 |
|
sess |
or well through pk |
| 16:52 |
|
Naros |
We generally haven't exposed the PK fields on the parent, only through the embedded id class itself |
| 16:52 |
|
sfisque |
the only non-transparency is when you need to limit on specific attributes of the CPK |
| 16:53 |
|
Naros |
right in this case, I want to get all WorkOrder entities where the CPK's workOrder number is like '%23%' for example |
| 16:53 |
|
sfisque |
so if you have PK(x,y,z) and you have a query/subquery where you need to limit on PK(y in (a,b,c)), or similar |
| 16:53 |
|
sess |
never used embeddedId so not sure if any class in generated |
| 16:53 |
|
sfisque |
then you have to explicitly call out the CPK in the query, otherwise, as sess indicates, it's just like any other pk |
| 16:54 |
|
Naros |
the hibernate metamodel generator doesn't give me a WorkOrderId_ class |
| 16:54 |
|
sess |
that was pretty much expected i guess |
| 16:54 |
|
sfisque |
is it annotated properly? |
| 16:54 |
|
sess |
or |
| 16:54 |
|
sess |
in your object using the ID, what type does the generated pk have? |
| 16:54 |
|
sess |
it has to have some type.. so it should be generated i guess |
| 16:54 |
|
Naros |
https://hibernate.atlassian.net/browse/METAGEN-105 |
| 16:54 |
|
Naros |
there's the code in question |
| 16:56 |
|
Naros |
hehe, see a typo in my post :/ |
| 16:56 |
|
sess |
uh embeddable + mappedsuperclass |
| 16:56 |
|
sess |
you sure have some weird stuff |
| 16:56 |
|
Naros |
Yep |
| 16:57 |
|
sess |
I really think WorkOrderId_ should be generated in this case |
| 16:57 |
|
sfisque |
never used @MappedSuperclass. what does that do? |
| 16:57 |
|
sess |
cant see any other way to access fields inside it |
| 16:57 |
|
sess |
you can inherit fields from it |
| 16:57 |
|
Naros |
@mappedSuperclass allows me to create another class that basically inherits those fields |
| 16:57 |
|
sfisque |
i'm guessing that confused hibernate |
| 16:57 |
|
sess |
for example, i have @MappedSuperclass AbstractEntity with the field @Id private Long id |
| 16:57 |
|
sess |
that most entities inherit from |
| 16:57 |
|
Naros |
Aye. |
| 16:58 |
|
Naros |
We do similar thngs but usually entitiy-specific |
| 16:58 |
|
sess |
though the metamodel IS generated from mapped superclasses aswell |
| 16:58 |
|
sess |
and i think it should be generated for embeddable objects aswell |
| 16:58 |
|
sess |
are you 100% sure it is not generated? |
| 16:58 |
|
Naros |
Those 2 fields are inherited in the WorkOrderDoodad, WorkOrderComponent, WorkOrderCost, and WorkOrderOperation PKs. |
| 16:58 |
|
Naros |
Yeah it isn't in my metamodel dir :E |
| 16:59 |
|
sfisque |
the other question i have, is the class generated for other examples that use the same construct (aka local failure, or systemic) |
| 16:59 |
|
Naros |
Nope |
| 16:59 |
|
Naros |
I see no Id classes generated anywhere. |
| 16:59 |
|
Naros |
across the entire domain landscape |
| 17:00 |
|
sfisque |
i'm guessing MSC and Embedded do not work together |
| 17:00 |
|
sess |
hmm one thought |
| 17:00 |
|
sfisque |
bug or feature |
| 17:00 |
|
sess |
Naros: is the embeddable object defined in your persistencexml? |
| 17:00 |
|
Naros |
No, we use the scanning features to locate entities and wire them in |
| 17:00 |
|
sfisque |
shouldnt matter. our persistence.xml is empty except for the junit version |
| 17:00 |
|
sess |
it could matter if he listed other entities |
| 17:00 |
|
sess |
but alright |
| 17:01 |
|
Naros |
So technically its my understanding I should have gotten a WorkOrderId_ |
| 17:01 |
|
sess |
I think so |
| 17:02 |
|
sess |
what happens if you remove @MappedSuperclass and try running the generator? |
| 17:02 |
|
Naros |
let me see. |
| 17:02 |
|
sess |
or create a new simple embeddable and see if it gets generated |
| 17:02 |
|
sess |
im getting off work now, so bbl |
| 17:03 |
|
Naros |
Doesn't make a difference. |
| 17:03 |
|
sfisque |
CPKs get generated in our project fine, but we do not use @MSC |
| 17:04 |
|
Naros |
Aye you're using a different generator tho i suspect. |
| 17:04 |
|
sfisque |
i just double checked for sanity and i see ***PK_ all over the place |
| 17:04 |
|
sfisque |
eclipselink |
| 17:04 |
|
Naros |
yah, using hibernate's |
| 17:04 |
|
sfisque |
but we also do not use @MappedSC |
| 17:05 |
|
sfisque |
simple embedded PKs |
| 17:05 |
|
Naros |
so then only solution is the non-typesafe string approach for these :E |
| 17:05 |
|
Naros |
Path<String> workOrderPath = workOrderRoot.get("id").get("workOrder"); |
| 17:05 |
|
Naros |
predicates.add(criteriaBuilder.equal(criteriaBuilder.lower(workOrderPath), workOrder.toLowerCase())); |
| 17:05 |
|
sfisque |
i have found trying to be "cute and slick" in entities only makes for headaches. i tend to let the entities be "VERY POJO" |
| 17:06 |
|
sfisque |
in the short term, you CAN move the strings into the entity as static finals, that way the code "feels" uniform, and then refactor when able |
| 17:06 |
|
Naros |
Aye, but even removing the inheritance stuff with MSC, seems hibernate's jpa metamodel generator isn't working with embeddedids |
| 17:06 |
|
sfisque |
that surprises me. it must be a regression because i've used hibernate to gen classes that had CPKs |
| 17:06 |
|
Naros |
right just expose the PK values on the entity itself |
| 17:06 |
|
sfisque |
in the past |
| 17:07 |
|
sfisque |
try rev'ing hibernate either up or down |
| 17:07 |
|
sfisque |
and see if it goes away |
| 17:07 |
|
Naros |
the metamodel gen right? |
| 17:07 |
|
sfisque |
yes |
| 17:07 |
|
sfisque |
no |
| 17:08 |
|
sfisque |
nm. i'm misremembering |
| 17:08 |
|
sfisque |
when i used hibernate, it was just before jpa2 came out |
| 17:08 |
|
sfisque |
so only used it for the model gen, not the metamotel gen |
| 17:08 |
|
sfisque |
***metamodel |
| 17:08 |
|
sfisque |
don't listen to me, i'm tired and distracted today :P |
| 17:09 |
|
sfisque |
either way, though sounds like you have either a "bug" or a "feature" |
| 17:10 |
|
Naros |
reverted from 1.3.0 all the way back to 1.0.0 and same output |
| 17:10 |
|
sfisque |
looks like hiberante MMgen has a hole |
| 17:10 |
|
Naros |
leave it to me to find it :E |
| 17:10 |
|
* sfisque |
hands Naros a hot cocao |
| 17:11 |
|
Naros |
I guess I can take your advice and expose the fields in the root entity and type-safe query from there. |
| 17:11 |
|
Naros |
and refactor when the MMgen is fixed. |
| 17:11 |
|
* sfisque |
nods |
| 17:12 |
|
sfisque |
if you look in the MM classes, they're just wrappers around lots of static finals string constructs |
| 17:13 |
|
sfisque |
all you'd be doing is relocating the missing reference temporarly until they get their butts in gear :P |
| 17:13 |
|
sfisque |
you "could" always hand craft the MMclass yourself, and tag it @Deprecated so it can get whacked if hibernate eventually produces it |
| 17:13 |
|
sfisque |
they're fairly trivial |
| 17:15 |
|
sfisque |
here's an example of one of ours gen'ed by eclipselink |
| 17:15 |
|
sfisque |
package com.k12.entity.metatypes; |
| 17:15 |
|
sfisque |
import javax.annotation.Generated; |
| 17:15 |
|
sfisque |
import javax.persistence.metamodel.SingularAttribute; |
| 17:15 |
|
sfisque |
import javax.persistence.metamodel.StaticMetamodel; |
| 17:15 |
|
sfisque |
@Generated(value="EclipseLink-2.4.2.v20130514-rNA", date="2013-10-29T16:22:59") |
| 17:15 |
|
sfisque |
@StaticMetamodel(StateCodePK.class) |
| 17:15 |
|
sfisque |
public class StateCodePK_ { |
| 17:15 |
|
sfisque |
public static volatile SingularAttribute<StateCodePK, Character> stateCodeCid; |
| 17:15 |
|
sfisque |
public static volatile SingularAttribute<StateCodePK, String> stateMachineCid; |
| 17:15 |
|
sfisque |
} |
| 17:16 |
|
sfisque |
basically just some SA bindings to make type safety happy :-) |
| 17:22 |
|
Naros |
Yep |
| 17:22 |
|
Naros |
That's true |
| 17:23 |
|
Naros |
Except I think the build process whacks the metamodel directory so I'll need to put these extra ones somewhere and copy them post generate sources |
| 17:24 |
|
Naros |
Could always just drop them in the src/main/java tree i suppose |
| 17:25 |
|
Naros |
now I really ponder why have these DAO classes :P |
| 17:26 |
|
Naros |
they pretty much abstracted hibernate-centric stuff |
| 17:30 |
|
Naros |
sfisque: this look right? |
| 17:30 |
|
Naros |
Root<WorkOrderId> workOrderIdRoot = criteriaQuery.from(WorkOrderId.class); |
| 17:30 |
|
Naros |
predicates.add(criteriaBuilder.equal(criteriaBuilder.lower(workOrderIdRoot.get(WorkOrderId_.workOrder)), workOrder.toLowerCase())); |
| 17:34 |
|
Naros |
answer appears to be no cause got java.lang.IllegalArgumentException: Not an entity: class com.setech.mrovelocityhub.persistence.maintenance.model.WorkOrder$WorkOrderId |
| 17:51 |
|
|
Nikitha joined ##javaee |
| 18:03 |
|
|
Nikitha left ##javaee |
| 18:03 |
|
Naros |
sfisque: I'm getting NPE using my manually created metamodel class |
| 18:03 |
|
Naros |
either it is not using it or im doing something wrong :) |
| 18:08 |
|
|
scripty joined ##javaee |
| 18:12 |
|
sfisque |
wierd |
| 18:12 |
|
sfisque |
sorry was afk |
| 18:12 |
|
Naros |
np. also found the mmgen issue |
| 18:12 |
|
sfisque |
typo in an annotation? |
| 18:12 |
|
Naros |
seems if the class is an inner class, it doesn't generate. |
| 18:12 |
|
sfisque |
OH |
| 18:12 |
|
Naros |
but if i move it outside as it's own class, it does find it |
| 18:12 |
|
sfisque |
for the love of all that is holy and java why would you have an inner class in an entity? |
| 18:13 |
|
Naros |
lol |
| 18:13 |
|
* sfisque |
slaps Naros with a huge noodle |
| 18:13 |
|
* Naros |
grins. |
| 18:13 |
|
Naros |
Looks like I have a bunch of refactoring to do :3 |
| 18:13 |
|
sfisque |
ROFL i bet |
| 18:13 |
|
Naros |
cause all my classes embed their PKs |
| 18:14 |
|
Naros |
but that aside, I still can't get my typed-safe query t owork |
| 18:14 |
|
sfisque |
right click, Refactor, move |
| 18:14 |
|
sfisque |
or i think its' Inner -> Outer |
| 18:14 |
|
sfisque |
have to double check.. one sec |
| 18:15 |
|
Naros |
criteriaBuilder.equal(workOrderRoot.get(WorkOrder_.id).get(WorkOrderId_.workOrder), workOrder); |
| 18:15 |
|
Naros |
Why does this give me NPE |
| 18:15 |
|
sfisque |
yah move Inner to Outer level in refactor |
| 18:15 |
|
sfisque |
did you instantiate the CB? |
| 18:15 |
|
Naros |
ofc :P |
| 18:16 |
|
Naros |
seems the first workOrderRoot.get(WorkOrder_.id) is what gives the NPE |
| 18:16 |
|
sfisque |
is the object "new" (aka no PK value?) |
| 18:16 |
|
Naros |
i dont follow |
| 18:16 |
|
sfisque |
aka x.id == null |
| 18:17 |
|
Naros |
I'm just trying to select work orders where workOrder.id.workOrder = value |
| 18:17 |
|
sfisque |
oh so workOrderRoot is a Root<x> gotcha |
| 18:17 |
|
Naros |
Yah |
| 18:17 |
|
sfisque |
have you gen'ed the MMs yet? |
| 18:18 |
|
Naros |
yep |
| 18:18 |
|
Naros |
id is being auto-gened now |
| 18:18 |
|
sfisque |
kk. not sure why you'd get an NPE. that seems "odd" |
| 18:18 |
|
Naros |
now it works :E |
| 18:19 |
|
Naros |
WTH |
| 18:19 |
|
sfisque |
caching? |
| 18:19 |
|
sfisque |
in the IDE? |
| 18:19 |
|
Naros |
shouldn't had been, i added some debugging changes to make sure it was building hte class and the new debug statements were written so IDK |
| 18:19 |
|
sfisque |
i know netbeans sometimes takea while to remove the red lines when i'm working with entities and MMs |
| 18:19 |
|
* Naros |
shrugs. |
| 18:19 |
|
Naros |
At least I got it workin |
| 18:20 |
|
sfisque |
gratz :-) |
| 18:20 |
|
Naros |
assumed i might had been doing something wrong |
| 18:20 |
|
sfisque |
yah, it wasnt apparent to me that your PKs were inner classes |
| 18:20 |
|
sfisque |
guess i skim read too fast |
| 18:20 |
|
Naros |
aye, I just had a hunch, let me try this and lo-and-behold. |
| 18:21 |
|
* Naros |
sighs |
| 18:21 |
|
sfisque |
aye. it's always good to "simplify" the problem. sometimes "getting cute" gets us into trouble |
| 18:21 |
|
Naros |
But like I like "cute" code :/ |
| 18:22 |
|
sfisque |
me too, but frameworks tend to barf on it |
| 18:22 |
|
Naros |
I just didn't care for having all these Pk classes in my model package |
| 18:22 |
|
Naros |
so I embedded them thinking that would be OK since most of the examples I saw long ago about Ids were public static classes in the entity. |
| 18:23 |
|
sfisque |
aye, you "can" do it, but it gives JPA some heartburn |
| 18:23 |
|
sfisque |
as you discovered |
| 18:23 |
|
Naros |
I am curious those if your mmgen creates an id |
| 18:23 |
|
Naros |
*curious tho |
| 18:23 |
|
Naros |
if your id was a pub static class |
| 18:24 |
|
Naros |
if it does, perhaps hibernate will fix it. if it doesn't, they may not and say that's expected behavior :o |
| 18:24 |
|
sfisque |
probably not. i think it has to do with the proxy'ing. i know putting static inner classes on EJBs can cause havoc |
| 18:24 |
|
Naros |
ah |
| 18:25 |
|
Naros |
Well, afternoon is shot to hell - haha |
| 18:25 |
|
sfisque |
i would posit that an inner class on any "managed object" is a poor design. managed objects are supposed to be anonymous and poolable. |
| 18:25 |
|
Naros |
good point, i just wasn't thinking, as i said mainly because most of the past examples I had encountered when learning ORM stuff had them as inner classes |
| 18:27 |
|
sfisque |
for instance, most people do not realize that "this" for an SLSB points to the pool, and not the instnace under the proxy. even doing something like method(){ this.otherMethod(); } doesnt guarantee you internal dispatch. that call "can" go to another instance of the SLSB in the pool if the container so wants |
| 18:27 |
|
sfisque |
that's why putting instance vars in SLSB that are not injected is a recipe for insanity |
| 18:28 |
|
|
nikita joined ##javaee |
| 18:28 |
|
nikita |
hii |
| 18:28 |
|
sfisque |
now imagine you have an inner class and you dispatch to OuterClass.this.method(). and you get a different reference. HELLO! now with static inner's it's different because you don't have that linkage, BUT i'm guessing instrumenting them "difficult" because of how they are bundled |
| 18:29 |
|
sfisque |
greets nikita |
| 18:29 |
|
nikita |
is there any book where i can start multi threading from grass |
| 18:30 |
|
sfisque |
javaeebot lucky doug lea concurrency java |
| 18:30 |
|
javaeebot |
sfisque: http://g.oswego.edu/ |
| 18:30 |
|
sfisque |
hrm |
| 18:30 |
|
sfisque |
javaeebot lucky java concurrency doug lea |
| 18:30 |
|
javaeebot |
sfisque: http://g.oswego.edu/ |
| 18:30 |
|
sfisque |
bah! |
| 18:31 |
|
sfisque |
sometimes reordering gives different ordering of results |
| 18:34 |
|
nikita |
is this a good start for java 1.7 |
| 18:36 |
|
Naros |
sfisque: very true :P |
| 18:36 |
|
Naros |
I guess I will "uncute" my code |
| 18:39 |
|
Naros |
I wonder whether they'll be any issues with public enums in classes |
| 18:39 |
|
Naros |
those shouldn't since they're mainly a type like String |
| 18:55 |
|
sfisque |
u can use enums in entities, we do |
| 18:56 |
|
Naros |
yah hibernate by itself had no issue, lokos like mmgen is ok with them too |
| 18:56 |
|
sfisque |
nikita yes. the only thing those guides might miss is the new concurrency stuff that has folded in lately. but its better to learn concurrency at the primitive level before diving in. |
| 18:57 |
|
Naros |
i like the jpa criteria api |
| 18:57 |
|
sfisque |
i do too. i hate embedding runtime strings in code |
| 18:57 |
|
Naros |
Im having a hard time now not just putting the em in my service classes |
| 18:58 |
|
Naros |
only hibernate specific thing i think i might end up using is unwrapping the typedQuery to a hibernate implementation to get scrollable results. |
| 18:59 |
|
sfisque |
you can't inject the EM? |
| 18:59 |
|
sfisque |
worse case, you "can" look it up via JNDI |
| 18:59 |
|
Naros |
no i mean, rather than injecting it into my DAO, I'm thinking of getting rid of the DAO and just inject right into the service tier. |
| 19:00 |
|
sfisque |
aye. i dislike dao pattern in an EE product. why abstract and already abstracted layer? |
| 19:00 |
|
sfisque |
unless you're doing something nutty like coalescing different datasources or such |
| 19:00 |
|
Naros |
aye or using vendor-specific stuff |
| 19:01 |
|
Naros |
which you can easily just use a wrapper |
| 19:01 |
|
Naros |
should you need to change the impl |
| 19:01 |
|
sfisque |
aye |
| 19:01 |
|
sfisque |
going afk. bbl. code strong! |
| 19:01 |
|
Naros |
see ya |
| 19:02 |
|
Naros |
curious whether anyone else here sees any point to the DAO tier if using straight jpa |
| 19:02 |
|
pdurbin |
hmm, I can't replicate my "Provider org.apache.xerces.jaxp.DocumentBuilderFactoryImpl not found" in a minimal maven app: https://github.com/pdurbin/oiosaml_on_glassfish |
| 19:02 |
|
semiosis |
Naros: no dao here |
| 19:22 |
|
pdurbin |
just posted: JavaOne 2013 Sessions - http://www.oracle.com/javaone/sessions/index.html |
| 19:52 |
|
|
Quest joined ##javaee |
| 19:55 |
|
SoniEx2 |
ok? |
| 19:55 |
|
Quest |
? |
| 19:55 |
|
SoniEx2 |
I'm not really active here... |
| 19:55 |
|
Quest |
I never require any one to either. |
| 19:56 |
|
Quest |
+v s also a priviledge to talk when channel is moderated |
| 19:57 |
|
SoniEx2 |
true... |
| 19:58 |
|
SoniEx2 |
but still... |
| 19:58 |
|
Quest |
Naros, I left a message but you were not here this morning. Just thought you would like to be interested in knowing. I did some research and found some code. It is possible to send / receive bare bytes via http (a good choice or bad is another story though. ). You might check logs if you want to. |
| 19:58 |
|
Naros |
k |
| 20:04 |
|
Quest |
Naros, found the log. line 947 http://pastebin.ca/2472645 application/octet-stream consists of "mere bytes" |
| 20:05 |
|
Naros |
correct, that is precisely what struts2 stream result-type does. |
| 20:05 |
|
Naros |
sets the content-type to that and streams the bytes to the browser. |
| 20:06 |
|
Naros |
But that's purely for the download side, not upload to my understanding. |
| 20:09 |
|
Quest |
that said in context to the server? |
| 20:09 |
|
Quest |
upload / download words i mean? |
| 20:09 |
|
Quest |
upload for the server and download for the server |
| 20:13 |
|
Naros |
I'm saying that makes sense to me downloading FROM the server. |
| 20:14 |
|
Naros |
But uploading, that won't work with form submission |
| 20:14 |
|
Naros |
because your requirement was not only send the file, but other information I thought. |
| 20:14 |
|
Naros |
unless you're planning on handling the upload via multiple requests. |
| 20:15 |
|
Quest |
the byte stream strategy can work for both upload to server and download from server. but in both cases, the client being a browser will make things difficult/impossible |
| 20:15 |
|
Quest |
i.e the client in my case was a mobile cell phone customized app to use http |
| 20:16 |
|
Naros |
Aye, on the browser side, you could set some session values, transmit form data and upload the file and marry them together. |
| 20:16 |
|
Naros |
but it would consist of transmitting the form data + file in multiple submissions |
| 20:16 |
|
Quest |
Naros, if you observer, its getting the file and ints again and again in the same stream |
| 20:17 |
|
Naros |
It shouldn't if you consider AJAX |
| 20:17 |
|
Quest |
strange. the post is deleted? can you see it |
| 20:17 |
|
Naros |
I didn't see anyhting in your pastebin |
| 20:17 |
|
Quest |
hm |
| 20:18 |
|
Naros |
my point is, if you go into gmail and try to send an email with an attachment, the attachment part is handling in ajax (separate request). |
| 20:18 |
|
Naros |
When you submit the email body, it gets uploaded to gmail and they marry the two together and generate a mim email with the attachment |
| 20:18 |
|
Naros |
essentially, you can do the same with your app |
| 20:18 |
|
pdurbin |
mime* |
| 20:18 |
|
Naros |
yah, typos are my specialty :P |
| 20:19 |
|
pdurbin |
:) |
| 20:20 |
|
Quest |
Naros, see line 61, 64, 108 http://pastie.org/8442323 |
| 20:20 |
|
|
SLovenberg joined ##javaee |
| 20:20 |
|
Quest |
Naros, yes. many other better options are there too. but was talking in that app scenario, where I was forced to implement the history again |
| 20:22 |
|
Naros |
that makes sense in a non-browser yes |
| 20:23 |
|
Naros |
you're using the HTTP protocol as a wrapper around a proprietary stream of data |
| 20:23 |
|
Quest |
yes |
| 20:23 |
|
Quest |
only because cell networks only allow http. |
| 20:24 |
|
Naros |
But as I indicated before, this is really no different than saying I am gonna run a server listening on port 80 and connect and send binary information |
| 20:24 |
|
Quest |
ya |
| 20:24 |
|
Naros |
Except in that case, I throw out all the HTTP jargon |
| 20:24 |
|
Quest |
ya |
| 20:25 |
|
Quest |
but the moral is , I got my ambiguity cleared and added another concept in my diary |
| 20:25 |
|
Naros |
Congrats. |
| 20:25 |
|
Quest |
that I am happy at. and shared it with others. |
| 20:25 |
|
Quest |
Thats whats makes me live. |
| 20:25 |
|
Quest |
thanks :) |
| 20:26 |
|
Naros |
what may get you tho is if you need to somehow manage stateful information :P |
| 20:26 |
|
Naros |
since HTTP by nature is stateless |
| 20:26 |
|
Quest |
I was really pissed of when I saw and was forced to implement their custome byte stream thing |
| 20:27 |
|
Naros |
HTTP is made stateful by using cookies or a unique session id sent on each request :P |
| 20:27 |
|
Quest |
Naros, yes. they send the params (sequenctional ints ) again and again |
| 20:27 |
|
Naros |
Yep |
| 20:28 |
|
Quest |
Naros, what i didnt got is how they jump to the next comming int by just doing utils.readInt(in); as in line 61, 64 and so on. |
| 20:28 |
|
Naros |
If you're sending lots of information in quick succession, HTTP is a poor choice imo. |
| 20:28 |
|
Quest |
http://pastie.org/8442323 |
| 20:28 |
|
Naros |
because open a socket connection is considered an expensive operation if the same endpoint is always used. |
| 20:29 |
|
Quest |
is it like that the pointer in the "in" stream goes one step ahead auto when utils.readInt(in); is called |
| 20:29 |
|
Naros |
yeah probably just maintained a position in the stream |
| 20:29 |
|
Naros |
and reads from the past point of read |
| 20:30 |
|
Quest |
line 1299 defines it. but i ddidnt got it though |
| 20:30 |
|
Naros |
if you look at InputStream, you'll notice it ahs a mark(int readlimit) method |
| 20:30 |
|
Naros |
http://docs.oracle.com/javase/7/docs/api/java/io/InputStream.html#mark(int) |
| 20:31 |
|
Naros |
scratch that, I misunderstood that api |
| 20:32 |
|
Quest |
you mean i should ignore the 3 lines above you stated? |
| 20:32 |
|
Naros |
yep |
| 20:32 |
|
Quest |
k |
| 20:32 |
|
Naros |
they have a variable called pos that maintains the read size |
| 20:32 |
|
Naros |
look in readFully() at 1342 |
| 20:33 |
|
Quest |
hm |
| 20:33 |
|
semiosis |
websockets |
| 20:33 |
|
Quest |
semiosis, not stable |
| 20:34 |
|
semiosis |
imho nothing web related is stable (except http) |
| 20:35 |
|
|
SLovenberg joined ##javaee |
| 20:36 |
|
Quest |
hm |
| 20:37 |
|
Quest |
Naros, quite low level customized protocol they made |
| 20:37 |
|
Naros |
its a typical byte stream, nothing more. |
| 20:37 |
|
Naros |
ones i've made in the past generally have some header that contains an identifier and size and variable payload. |
| 20:38 |
|
Naros |
read header, determine operation & payload size, read payload, perform operation. |
| 20:42 |
|
pdurbin |
crap. my xerces/glassfish/maven problem is back again. everything was working fine for about an hour |
| 20:42 |
|
Naros |
that usually happens right after a clean & rebuild :/ |
| 20:44 |
|
|
SLovenberg joined ##javaee |
| 20:44 |
|
pdurbin |
Naros: yeah. exactly |
| 20:45 |
|
pdurbin |
well, at least I got a whole hour of coding in today :/ |
| 20:46 |
|
Quest |
Naros, last question before i wave. string 'a' can be converted to an int. now lets say 'a' = 58 b 59 c 60 . Now in a stream, when it reads ints, how does it knows that its a,b,c like ,58,59,60 and not 585,960 ? |
| 20:47 |
|
Naros |
lol i've been rewriting my util helpers to JPA criteria, you can do that fo rme :P |
| 20:47 |
|
Naros |
because both are basically 8 bytes :P |
| 20:48 |
|
Quest |
8 or 4 ? |
| 20:48 |
|
Naros |
oops, 8 bits |
| 20:48 |
|
Naros |
a char and int are generally interchangable. |
| 20:48 |
|
Quest |
a char is 8bytes |
| 20:49 |
|
Quest |
the line 1300 stats byte[] intBytes = new byte[4]; though |
| 20:49 |
|
Naros |
sorry, getting confused with C++ |
| 20:49 |
|
Naros |
sizeof(char) = 1 in C++ :P |
| 20:50 |
|
Naros |
Anyway, 'A' is 65 |
| 20:50 |
|
Naros |
it's the ascii character code, integer representation. |
| 20:51 |
|
Naros |
if you write Characer.toChars(65); it would print out an 'A' |
| 20:51 |
|
Quest |
ya but why http://pastie.org/8442323#1300 a char is 8bytes. why the line 1300 stats byte[] intBytes = new byte[4]; though |
| 20:52 |
|
Quest |
'A' is actually an 8byte char |
| 20:52 |
|
Quest |
why the [] isnt new byte[8] |
| 20:54 |
|
Naros |
the protocol transmits integers using 4 bytes. it then uses bit arithmetic to create the int inside byteArrayToInt() |
| 20:54 |
|
Naros |
none of this has anything to do with char :P |
| 20:55 |
|
Naros |
look at like 1441 |
| 20:55 |
|
Quest |
"the protocol transmits integers using 4 bytes" .. |
| 20:55 |
|
Naros |
Do you see line 1444 where they compute the shift value using * 8 :P |
| 20:56 |
|
Naros |
Programming languages often represent a value using X bytes but when they get sent over a protocol, they're often packed into a smaller value to avoid bloating the protocol and sending lots of unnecessary bytes. |
| 20:57 |
|
Quest |
1441 yes |
| 20:57 |
|
Quest |
1444 yes yes |
| 20:57 |
|
Naros |
So for example, lets say I wanted to represent in my code an int, but when I send it, I send it as 1 byte because I know it will never exceed 255. |
| 20:57 |
|
Naros |
this way I minimize bandwidth |
| 20:58 |
|
Naros |
This type of stuff is fairly common in custom protocols. |
| 20:58 |
|
Quest |
hm |
| 20:58 |
|
Quest |
I would have to read details to get that. where can i read? |
| 20:59 |
|
semiosis |
can't you just gzip it? ;) |
| 20:59 |
|
Quest |
semiosis, we already do that |
| 20:59 |
|
semiosis |
heh ok |
| 20:59 |
|
Naros |
It's just something I learned from my last job Quest. |
| 21:00 |
|
semiosis |
also curious whats the benefit of maintaining a custom protocol when there's things like protobuf, thrift, msgpack, 0mq, ..... the list goes on |
| 21:00 |
|
Quest |
you learned via practical experience? but there must be some textual readings too. |
| 21:00 |
|
Naros |
We needed to send as small of packets as possible between the game server & game clients, so we often did a computation to pack data on server side and unpacked it on the client end so that for lower-end internet connections, the game was still fairly playable. |
| 21:00 |
|
Quest |
Naros, semiosis I wish there would be webpage gzip too. so we load the pages and surf faster |
| 21:01 |
|
semiosis |
Quest: ??? |
| 21:01 |
|
Quest |
semiosis, using gzip for browsing |
| 21:01 |
|
semiosis |
lots of web servers can and do send gzipped data |
| 21:01 |
|
Naros |
Yep |
| 21:01 |
|
Quest |
but browsers only load html/text |
| 21:01 |
|
Naros |
primarily for the same reason as I started just above, for those low-end bandwidth connections and to also minimize overall bandwidth usage. |
| 21:01 |
|
Quest |
hm |
| 21:01 |
|
semiosis |
the browser & server negotiate (using the Accept header) and can agree to gzip the http payload |
| 21:02 |
|
Naros |
if the server never sees the browser supports gzip, it just sends it plain text/html |
| 21:02 |
|
semiosis |
Accept-Encoding specifically |
| 21:02 |
|
Quest |
semiosis, hm. so we CAN load an html page via gzip sent from server? |
| 21:02 |
|
Naros |
if your client supports decoding the gzip stream, sure. |
| 21:02 |
|
Naros |
and if the server supports sending gzip :P |
| 21:03 |
|
Quest |
Naros, firefox, chrome, IE , opera, Konqueror does i gues? |
| 21:03 |
|
semiosis |
most likely |
| 21:03 |
|
Naros |
Most do, but having a fallback to non-gzip is always good practice. |
| 21:03 |
|
Quest |
Naros, iam pretty hoping and near sure that tomcat and apache httpd does too |
| 21:04 |
|
Quest |
Naros, semiosis then i guess its a good practice and should be adopted? |
| 21:04 |
|
Quest |
always adopted? |
| 21:04 |
|
Naros |
Quest: Apache does if you enable mod_deflate iirc. |
| 21:04 |
|
Quest |
hm |
| 21:04 |
|
Naros |
Whether that mod is enabled by default, I'd have to check |
| 21:05 |
|
Quest |
Naros, so I would just configure the mod and no need to recode the app |
| 21:05 |
|
Quest |
? |
| 21:05 |
|
Naros |
http://httpd.apache.org/docs/2.2/mod/mod_deflate.html |
| 21:05 |
|
Naros |
"The mod_deflate module provides the DEFLATE output filter that allows output from your server to be compressed before being sent to the client over the network." |
| 21:05 |
|
Quest |
if yes, greate |
| 21:05 |
|
Naros |
Yes, it should |
| 21:06 |
|
Naros |
For tomcat, I suspect you might need to apply in your web.xml some filter to do this |
| 21:06 |
|
Quest |
it will make a lot of CPU usage of server and somewhat of client too but would benifit in consuming less bandwidth . for both client and server |
| 21:07 |
|
Naros |
cpu usage will be overall negligable. |
| 21:07 |
|
Quest |
negligable? well its a full compression of everything. |
| 21:07 |
|
Quest |
extra cpu usage will/ should occur |
| 21:08 |
|
Naros |
naturally, but until you profile it - consider it negligable :P |
| 21:08 |
|
Quest |
:) ya |
| 21:08 |
|
Naros |
*negligible |
| 21:08 |
|
Quest |
thanks .. pleaseure talking |
| 21:09 |
|
Quest |
you guys are opposite of #java |
| 21:11 |
|
|
Quest left ##javaee |
| 21:15 |
|
|
SLovenberg joined ##javaee |
| 21:53 |
|
sess |
gzipping responses probably only takes a millisecond |
| 21:53 |
|
sfisque |
so, design question regarding nosql db's (hadoop, mongodb, etc.). if these db's are name/value pools, is there a way to represent hierarchical data without having to "fake it" by hiding FK's (name keys) in the value payload? |
| 21:54 |
|
sess |
the whole point of nosql databases is being able to store lists in single "columns" |
| 21:54 |
|
sess |
i.e perfect for hierarcihical data |
| 21:55 |
|
sfisque |
that does not explain the mechanic though. how do you represent hierarchy in a flat address space? |
| 21:55 |
|
sess |
flat? |
| 21:56 |
|
sfisque |
name, value pairs are a 2d vector space. flat |
| 21:56 |
|
sess |
but what if value can be a complex object? |
| 21:56 |
|
sfisque |
that's what i figured, you have to cook the data model |
| 21:56 |
|
sfisque |
you dont get hierarchy for free, you have to bake it into the values |
| 21:57 |
|
sess |
what do you mean "bake" |
| 21:57 |
|
sfisque |
i was wondering if there was some "other" mechanic that these repo's used to handle that |
| 21:58 |
|
sfisque |
bake - as in it's not native. i can build a hierarchical model in an RDBMS without putting any data into it. in order for a flat data space to have hierarchy, you have to either bake the data (encode the structure into the data itself) or the product has to provide some "mechanic" to enforce it |
| 21:59 |
|
sess |
as i understand, nosql is schemaless |
| 22:00 |
|
sfisque |
aye. i was just curious if there was a "typical" feature of the popular repos to "handle hierarchy", or if it's implementor driven |
| 22:00 |
|
sfisque |
your feedback would lead me to believe it is the latter |
| 22:01 |
|
sfisque |
i'm thinking of looking into nosql, but having to bake the data makes it unusable for anything i'm working on. i don't have that kind of driver in my current data. |
| 22:01 |
|
sess |
ive never used any nosql db, but it's supposed to be much better for hierarchicál structures than normal rdbms |
| 22:02 |
|
sess |
im not quite sure what you mean |
| 22:02 |
|
sess |
you dont have to do anything special with your data |
| 22:03 |
|
sess |
sfisque: think of it like being able to store a complex json object |
| 22:04 |
|
sess |
with a single command |
| 22:04 |
|
sess |
document.put('somekey', complexobject) |
| 22:04 |
|
sfisque |
denormalized data means every query is a full crawl, no optimizations, no clear way to slice/dice unless i "roll my own" which i have better things to do than reinvent the wheel. i dont have any data that is purely consumptive |
| 22:04 |
|
sfisque |
sess i understand name, value |
| 22:05 |
|
sfisque |
the issue is, i dont have "Resource" type data. it's all very crud centric and structured, thus sql is the proper solution. |
| 22:06 |
|
sfisque |
i was thinking if there were "sql-esque" solutions that can be leveraged on nosql, i'd take the dive |
| 22:06 |
|
sess |
perhaps, i never tried nosql so im not suire where it's appropriate |
| 22:06 |
|
sfisque |
if for no better reason, than to try out the tech |
| 22:06 |
|
sfisque |
aye, same, i'm basically "looking for an excuse" but i dont want to end up with a mess :P |
| 22:07 |
|
sess |
it's probably good enough for most if not any normal usages |
| 22:07 |
|
sess |
if the application doesnt have special demands |
| 22:07 |
|
sfisque |
and requires heavy relational querying |
| 22:07 |
|
sfisque |
***doesnt |
| 22:07 |
|
sess |
it's supposed to be really fast |
| 22:08 |
|
sfisque |
well yah. as long as it has a good hashing algorithm, it should be |
| 22:08 |
|
sfisque |
a linear sweep is hella potentially faster than building explain plans on the fly, if you have optimized hashing algorithms |
| 22:10 |
|
sfisque |
but i'd also assume doing somethign like get( "%some_substring%" ), whould be attrocious unless you have a custom optimizer to build indexes that you can then fetch and iterate though |
| 22:11 |
|
sess |
frankly i have no idea how conditional querying is done |
| 22:11 |
|
sess |
cant imagine |
| 22:11 |
|
sfisque |
basically have scheduled agents sweep and build indexes and store them at predefined keys |
| 22:11 |
|
sfisque |
that's the only way i can see it having any performance |
| 22:11 |
|
sfisque |
or it might not be a real concept. what i'm envisioning might be considered an example of "why would you even do that" |
| 22:12 |
|
sfisque |
/shrug |
| 22:12 |
|
sfisque |
i'll have to read up some more then |
| 22:14 |
|
sess |
i dislike the idea of lacking ACID |
| 22:14 |
|
sess |
and going without schema |
| 22:14 |
|
sess |
maybe its just something for really big data |
| 22:21 |
|
sfisque |
yeah.. i found a page discusses the question of "why". interesting . it proves useful i'll link it |
| 22:21 |
|
sfisque |
** if it proves... |
| 22:21 |
|
zoot |
rdf/sparql has ACID |
| 22:22 |
|
zoot |
and also a usecase ;) |
| 22:22 |
|
sfisque |
aye. the real driver for me is "what aspect of data would promote nosql over sql" that's what i'm trying to discover |
| 22:23 |
|
sfisque |
i can see the value if you're just storing for direct retrieval, but what about culling and querying |
| 22:23 |
|
zoot |
most other variants of nosql is just another filesystem.. |
| 22:30 |
|
sfisque |
so what i'm reading nosql is not a replacement per se of an RDBMS, but allows offloading some aspects that don't quite fit well for an application. so you would still use a traditional rdbms for data that is highly relational, and use nosql for stuff that is more "amorphous" or has a slimmer data model and requires fast "flat" retrieival |
| 22:31 |
|
sfisque |
so it would become part of a heterogenous data model where you use it for "certain types" of data, but not all of your data needs. |
| 22:31 |
|
sfisque |
that makes sense to me |
| 22:31 |
|
sfisque |
http://highscalability.com/blog/2010/12/6/what-the-heck-are-you-actually-using-nosql-for.html |
| 22:32 |
|
pdurbin |
sfisque: maybe your facebook killer could be built on nosql |
| 22:32 |
|
pdurbin |
built on neo4j |
| 22:33 |
|
sfisque |
i'm thinking nosql would definitely be part of the mix. basically have "profile info" in sql, and "content" in nosql |
| 22:33 |
|
|
Naros joined ##javaee |
| 22:34 |
|
sess |
sounds reasonable sfisque |
| 22:34 |
|
sfisque |
because being able to "cull" based on profile (who's in my list of "friends" or "subscribers" is relational. but the list of "feed status bits" would be purely name value paired |
| 22:34 |
|
sess |
kind of like using lucene |
| 22:34 |
|
sfisque |
aye |
| 22:34 |
|
sess |
which is a bitch at points |
| 22:35 |
|
sfisque |
that link was pretty good. at the end it even lists a bunch of "bad uses for nosql" |
| 22:35 |
|
sfisque |
which helps me understand better |
| 22:36 |
|
sfisque |
a neat concept would be offloading "blobs". have a table that has "meta data" and then a "key" into nosql for some arbitrary content like "description" or "pdf version" that could badly bloat a schema |
| 22:37 |
|
sfisque |
so the relational parts are in sql, and then the "byte stream stuffs" would be key/value referenced |
| 22:38 |
|
sfisque |
the referrence could even be a rest endpoint because we don't care about joining against it, just retrieval or update |
| 22:38 |
|
sfisque |
it's starting to make sense now |
| 23:27 |
|
|
whartung joined ##javaee |
| 23:32 |
|
pdurbin |
sfisque: you might be interested in this: Using MongoDB in a Java Enterprise Application [BOF7857] https://oracleus.activeevents.com/2013/connect/sessionDetail.ww?SESSION_ID=7857 |
| 23:33 |
|
pdurbin |
given my people in my department, in the office two doors down |
| 23:33 |
|
sfisque |
cool |
| 23:34 |
|
acuzio |
I am still making sense of the difference between K/V stores and nosql |
| 23:38 |
|
semiosis |
javaeebot: lucky spring data mongodb |
| 23:38 |
|
javaeebot |
semiosis: http://projects.spring.io/spring-data-mongodb/ |
| 23:40 |
|
tjsnell |
acuzio: give up, it's above your pay grade |
| 23:42 |
|
whartung |
nosql covers a larger swath of database tech, including K/V stores |
| 23:43 |
|
acuzio |
tjsnell: i agree - i think i need more words added to it like Cloud, Virtual, Horizontal computing |
| 23:43 |
|
tjsnell |
yep |
| 23:44 |
|
tjsnell |
and webscale |
| 23:44 |
|
tjsnell |
but that's a given |
| 23:44 |
|
acuzio |
whartung: K/V stores with a way to access it via JSON, XML, ProtoBuffers, RPC-API's ; what else is it ? Many of them scale across and replicate but so what |
| 23:44 |
|
whartung |
not at NoSQL dbs are KV stores. |
| 23:44 |
|
whartung |
*all |
| 23:47 |
|
acuzio |
yeah some of them are "document" stores |
| 23:48 |
|
whartung |
si |
| 23:48 |
|
whartung |
Couch is more a document store than a K/V store, for example |
| 23:48 |
|
acuzio |
si |
| 23:48 |
|
acuzio |
It used to be called Blobs |
| 23:49 |
|
whartung |
I find couch fascinating, shame its so slow |
| 23:49 |
|
whartung |
there's also the columnar databases |
| 23:52 |
|
pdurbin |
and graph databases (neo4j) |
| 23:53 |
|
acuzio |
graph databases are interesting - |
| 23:53 |
|
pdurbin |
wicket efficient for certain use cases |
| 23:53 |
|
pdurbin |
wicked* |
| 23:54 |
|
acuzio |
but hey in my day - when you wanted to store unorganized stuff in a DB - we used XML or blob |
| 23:56 |
|
pdurbin |
sfisque: heh, you and I were chatting during my note taking for that mongo talk: http://irclog.greptilian.com/javaee/2013-09-24#i_24881 |
| 23:56 |
|
pdurbin |
cheeser also gave a mongo talk at java one. I heard it was good. he works for the makers of mongo |
| 23:56 |
|
acuzio |
in tjsnell;s day - they didnt have DB's |
| 23:56 |
|
acuzio |
yeah |
| 23:57 |
|
tjsnell |
Sybase |
| 23:57 |
|
tjsnell |
it should be on the web as of today |
| 23:57 |
|
tjsnell |
J1 talks went up |
| 23:57 |
|
acuzio |
sybase was big in the finance sector around 2003/2005 - it was speedier than orcl |
| 23:58 |
|
acuzio |
its becoming the cool thing to be at J1 again pity i was just beginning to sound like an old pro |
| 23:59 |
|
sfisque |
really? from what i've seen, sybase started to slide in popularity around 2000 |
