| Time |
S |
Nick |
Message |
| 00:01 |
|
whartung |
this reminds me of an old mac commercial |
| 00:01 |
|
whartung |
2 engineers, late in the day, thumbing through manuals |
| 00:01 |
|
whartung |
a lady walk up |
| 00:01 |
|
whartung |
"You two have been at this all week, what are you doing?" |
| 00:01 |
|
whartung |
"Making it easier to use!" |
| 00:02 |
|
whartung |
they're struggling to install Windows on a PC |
| 00:07 |
|
whartung |
weld apparently has no logging... |
| 00:11 |
|
whartung |
back to a PropertyChangeListener |
| 00:15 |
|
sfisque |
i remember that commercial |
| 00:16 |
|
whartung |
I might just use a JMS topic |
| 00:16 |
|
sfisque |
going afk for a bit. i have a general question that hopefully someone can chime in on. is there an event or interceptor that i can use to capture the moment jaas return a valid principal/credential? |
| 00:17 |
|
whartung |
return from where? |
| 00:17 |
|
sfisque |
that will definitely work because mdb's are top level ejbs so you'll have the proper context for asynch processing |
| 00:18 |
|
sfisque |
from the container. as in the container intercepts the request, does the jaas "present the login screen", processes the form input, performs the redirect back to the original request. i want to capture the fact that the auth happened because i have "things that happen" based on user stuff |
| 00:18 |
|
whartung |
ah -- dunno if that lifecycle is really exposed |
| 00:18 |
|
whartung |
or if JAAS is the way to even do it today |
| 00:19 |
|
sajjadg |
Apache Shiro :-) |
| 00:19 |
|
whartung |
JEE constainers implement a couple of other JSRs for authentication nowadays |
| 00:22 |
|
sfisque |
jaas works fine. no need to reinvent the wheel AFAIC. i have it working but i have to repoll for the user if it's null rather than just setting it once and reacting appropriately |
| 00:22 |
|
whartung |
I was never able to get JAAS to work with a web form. |
| 00:22 |
|
whartung |
we use our own forms and programmatic login on GF |
| 00:43 |
|
SoniEx2 |
which one is faster: `metadata[recipe[index - 1]]` or a `last` parameter? |
| 00:43 |
|
whartung |
eh? |
| 00:44 |
|
SoniEx2 |
uhh wait |
| 00:44 |
|
SoniEx2 |
https://github.com/SoniEx2/NewEnderMoney/blob/1.7.2/common/com/github/soniex2/endermoney/core/item/EnderCoin.java#L46 |
| 00:45 |
|
SoniEx2 |
see that `metadata[recipe[index - 1]]`? |
| 00:45 |
|
whartung |
yea |
| 00:45 |
|
SoniEx2 |
would it be faster to replace it with a `last` parameter? |
| 00:45 |
|
whartung |
I don't know what "last" is |
| 00:46 |
|
SoniEx2 |
uhh wait |
| 00:46 |
|
SoniEx2 |
basically I would add `, int last` to the method |
| 00:46 |
|
SoniEx2 |
and that would be v |
| 00:47 |
|
whartung |
well in this case, v changes as it's part of the loop |
| 00:48 |
|
whartung |
you could pull the metadata[recipe[index - 1]] out of the loop and then use that directly, but odds are really quite high the compiler is doing this for you already, so I wouldn't worry about it |
| 00:48 |
|
whartung |
whatever is slowing you down, this isn't it. |
| 00:48 |
|
SoniEx2 |
uhh well |
| 00:49 |
|
SoniEx2 |
metadata[recipe[index - 1]] would be like calling the method with the last v? |
| 00:49 |
|
SoniEx2 |
https://github.com/SoniEx2/NewEnderMoney/blob/1.7.2/common/com/github/soniex2/endermoney/core/item/EnderCoin.java#L45 |
| 00:49 |
|
whartung |
no, these are arrays - no methods involved here at all |
| 00:49 |
|
SoniEx2 |
https://github.com/SoniEx2/NewEnderMoney/blob/1.7.2/common/com/github/soniex2/endermoney/core/item/EnderCoin.java#L48 |
| 00:50 |
|
SoniEx2 |
it calls itself up to 9 times |
| 00:51 |
|
whartung |
but a) since neither metadata or recipe are marked as volatile, this is effectively a constant expression, so the compiler (or JIT) is quite likely moving this out of the loop for you already, and thus only does the calculation once |
| 00:51 |
|
whartung |
but in the end, these are simple array deferencecs |
| 00:51 |
|
SoniEx2 |
still |
| 00:52 |
|
whartung |
still what? these are cheap..cheap cheap…super cheap…sub nanosecond exections. |
| 00:52 |
|
SoniEx2 |
when the method gets called in the loop, recipe[index - 1] from the new method call = old x |
| 00:52 |
|
SoniEx2 |
and v = metadata[x] |
| 00:52 |
|
whartung |
I can guarantee you that you have FAR FAR worse performance problems that need you attention than this expression. |
| 00:52 |
|
SoniEx2 |
well yeah |
| 00:52 |
|
SoniEx2 |
it's my doTrade method :P |
| 00:53 |
|
whartung |
ok |
| 00:53 |
|
SoniEx2 |
but I have to update it to 1.7 so... |
| 00:53 |
|
whartung |
so nothing |
| 00:54 |
|
whartung |
this isn't; relevant to that either |
| 00:54 |
|
SoniEx2 |
you want me to change CoinRecipe? |
| 00:54 |
|
whartung |
regarding this specific expression, I don't want you to change anything |
| 00:54 |
|
whartung |
it's fine as it is |
| 00:55 |
|
SoniEx2 |
uhh ok |
| 00:55 |
|
SoniEx2 |
what am I doing wrong? |
| 00:55 |
|
whartung |
what do you mean? |
| 00:55 |
|
SoniEx2 |
well you said I have performance problems? |
| 00:56 |
|
whartung |
no, YOU said you had performance problems |
| 00:56 |
|
SoniEx2 |
no |
| 00:56 |
|
SoniEx2 |
I'm just trying to know which one is faster |
| 00:56 |
|
whartung |
"which is faster, x or y" implies "performance problems" |
| 00:56 |
|
whartung |
the expression is derefecenceing two arrays, repeatedly in a loop |
| 00:57 |
|
SoniEx2 |
I don't think the JIT is smart enough to replace metadata[recipe[index - 1]] with the v from the caller... |
| 00:57 |
|
whartung |
naively, since this is a constant expression, it would be "better" to move that calculation out of the loop |
| 00:57 |
|
whartung |
but the compiler/jit is "smart enough" to do that for you. |
| 00:57 |
|
whartung |
and in any case |
| 00:57 |
|
whartung |
until it shows up on the top of your profiler, it's not even worth spending a nanosecond of thought on in consideration. |
| 00:58 |
|
SoniEx2 |
moving it out of the loop... well that wouldn't use the v so it would do index - 1, then index from recipe, then index from metadata |
| 00:58 |
|
whartung |
metadata[recipe[index - 1]] doesn't use 'v' a tall |
| 00:58 |
|
SoniEx2 |
I think that's slower than just adding the v to the param list |
| 00:58 |
|
whartung |
at all |
| 00:58 |
|
SoniEx2 |
well yeah that's what I said... |
| 00:58 |
|
whartung |
metadata is constant, recipe is constant, index is constant |
| 00:58 |
|
whartung |
-1 is constant |
| 00:59 |
|
whartung |
for the duration of the loop |
| 00:59 |
|
whartung |
so if you don't want to do the math repeatedly, move it out of the loop |
| 00:59 |
|
SoniEx2 |
well yeah |
| 00:59 |
|
SoniEx2 |
but it still requires a single calculation |
| 00:59 |
|
SoniEx2 |
that I could avoid by adding the v |
| 00:59 |
|
whartung |
q = metadata[recipe[index - 1]]; if ((index == 0 || v >= q) && v <= left) { |
| 01:00 |
|
SoniEx2 |
I would name it `last` |
| 01:00 |
|
SoniEx2 |
not q |
| 01:00 |
|
whartung |
you don't want to make the CALLER calculate this, as it's not their concern. |
| 01:01 |
|
SoniEx2 |
no it's already calculated by the caller |
| 01:01 |
|
whartung |
this sentinel value is a problem for the registerRecipes function, the caller can care less |
| 01:02 |
|
SoniEx2 |
also it's recursive up to 9 levels because it's kind of a loop... |
| 01:02 |
|
SoniEx2 |
a loop within a loop within a loop or something |
| 01:03 |
|
SoniEx2 |
+ I ported the thing from Lua which doesn't exactly help... |
| 01:09 |
|
|
kotten_ joined ##javaee |
| 01:11 |
|
sfisque |
ooooh jboss has an MBean exposed in jaas :-) |
| 01:12 |
|
whartung |
nice |
| 01:12 |
|
whartung |
not surprising, know jboss structure |
| 01:12 |
|
whartung |
MBeans were their hammer |
| 01:12 |
|
SoniEx2 |
sfisque: could you help me? |
| 01:12 |
|
sfisque |
jboss has a very deep and effective jaas history |
| 01:12 |
|
sfisque |
i've used their jaas since v4 |
| 01:12 |
|
whartung |
how do you get jaas to work with a web form? |
| 01:13 |
|
sfisque |
web.xml - configure for form. tie security to the container security context (jboss-web.xml attribute) |
| 01:13 |
|
whartung |
so basically a jaas realm |
| 01:13 |
|
sfisque |
container handles everything. all u do is configure the driver(s), supply a login form, done |
| 01:13 |
|
sfisque |
yes |
| 01:14 |
|
whartung |
ok |
| 01:14 |
|
whartung |
we implemented a GF realm |
| 01:14 |
|
whartung |
but still use our own forms |
| 01:14 |
|
sfisque |
u always supply a form. unless you use "basic"… bleh |
| 01:14 |
|
sfisque |
the difference is, who handles the marshalling and interception |
| 01:15 |
|
sfisque |
jaas == container does the heavy lifting |
| 01:15 |
|
sfisque |
the nice part of jaas, the security is exposed to the ejb layer. you can annotate methods to check for principal or roles and have them throw exceptions if the caller does not have the correct authorizations |
| 01:15 |
|
whartung |
we have about 99 auth type sna what not that we support in our app. |
| 01:15 |
|
whartung |
we can still do that with programmatic login |
| 01:15 |
|
whartung |
all that applies |
| 01:16 |
|
whartung |
just a matter of how to register the principal with the container |
| 01:16 |
|
sfisque |
aye. i just like leveraging the container whenever possible because why should i write code needlessly :P |
| 01:16 |
|
SoniEx2 |
sfisque: help me? |
| 01:17 |
|
sfisque |
i would but i'm tight on time. buddies coming over in a bit for some table top gaming :-D |
| 01:17 |
|
SoniEx2 |
ok |
| 01:17 |
|
whartung |
what are you playing? |
| 01:18 |
|
sfisque |
finishing up a pathfinder (meh) story arc, then next week we're switching to a classic traveller campaign (i'm running that) for a little change in scenery |
| 01:18 |
|
* whartung |
old school traveller buff |
| 01:19 |
|
whartung |
I have pretty much everything. |
| 01:19 |
|
whartung |
I bought my boxed set in '79 |
| 01:19 |
|
whartung |
all of that stuff, all of the MT stuff, all of the TNE stuff. |
| 01:20 |
|
whartung |
Azhanit High Lightning, Fifth Frontier War |
| 01:20 |
|
whartung |
Batller Rider, Brilliant Lances…I have a full set of the original JTAS |
| 01:20 |
|
sfisque |
i liked some of the MT rules changes, but i disliked the canon stuff. this campaign is classic with some "sci-fantasy" thrown in |
| 01:20 |
|
sfisque |
nice |
| 01:20 |
|
sfisque |
i'm rebuilding my collection of classic stuff. sold off a ton years ago, now regret it |
| 01:21 |
|
sfisque |
still have my AHL deck plans though. couldnt part with them |
| 01:21 |
|
whartung |
obviously you can get all electronic |
| 01:21 |
|
sfisque |
aye but it's not the same |
| 01:21 |
|
whartung |
yea I somehow lost the actual Supplement 5 (I think) that came with the game |
| 01:21 |
|
whartung |
but I have the maps and counters |
| 01:22 |
|
whartung |
I have a acknowledgement in Fire Fusion and Steel :D |
| 01:22 |
|
sfisque |
cool |
| 01:22 |
|
whartung |
I'z famous |
| 01:22 |
|
whartung |
send money in accolades |
| 01:22 |
|
* sfisque |
tosses whartung an imperial credit |
| 01:23 |
|
whartung |
*clink* |
| 01:24 |
|
whartung |
I watch T5 with interest, but it looks like a continuing disaster, so I lost interest |
| 01:25 |
|
whartung |
I haven't been to the traveller forums in months |
| 01:25 |
|
sfisque |
from what i saw it's pretty much MT and TNE folded together with the classic canon |
| 01:26 |
|
whartung |
so is JNDI stable @Startup? |
| 01:26 |
|
sfisque |
yes JNDI is container wide |
| 01:26 |
|
sfisque |
unless the app itself does bindings |
| 01:26 |
|
sfisque |
which is rare unless you're implementing your own directory context |
| 01:27 |
|
sfisque |
JNDI has to be stable. it's how you lookup EJB's (annotations are just shorthand for looking them up) |
| 01:28 |
|
sfisque |
bingo, all i have to do is build a skeletal loginModule and have ti fire last. basically dummy up a loginModule as an event handler |
| 01:29 |
|
sfisque |
i should join the jaas jcp, that api could use some love |
| 01:29 |
|
sfisque |
there really should be lifecycle handlers and other goodies |
| 01:30 |
|
sfisque |
so for now i keep doing what i'm doing, but i now have a refactor target :-) |
| 01:31 |
|
whartung |
dunno if jaas is the place for that |
| 01:31 |
|
whartung |
\o/ mdb worked |
| 01:31 |
|
sfisque |
why not? i mean tehre is a life cycle (pre-auth, authent, authorize, de-auth (logout) |
| 01:31 |
|
sfisque |
so there should be some event handling exposed to the apps |
| 01:31 |
|
sfisque |
NICE |
| 01:32 |
|
whartung |
but in the container they have those other security JSRs also |
| 01:32 |
|
sfisque |
but how many of them are ready for primetime in EE7? i mean jaas has been around F O R E V E R |
| 01:33 |
|
whartung |
Dunno -- they're all already IN 6 |
| 01:33 |
|
sfisque |
for regular web pages (not rest or ws) |
| 01:33 |
|
sfisque |
? |
| 01:34 |
|
sfisque |
educate me. i'm open to newer tech if it's stable :-D |
| 01:34 |
|
sfisque |
if its off the shelf in the container, even better |
| 01:34 |
|
whartung |
I'm so bummed this CDI stuff didn't work |
| 01:35 |
|
sfisque |
CDI has issues in EE6/7. hopefully in EE8 they'll finish removing ManagedBean and the other cruft and work out the EJB/CDI linkage grey areas |
| 01:36 |
|
whartung |
I'll have to hunt down the jars |
| 01:36 |
|
whartung |
jsr |
| 01:36 |
|
whartung |
it might have to do with standardizing programmatic security |
| 01:37 |
|
sfisque |
looking at the EE7 jsr list. none are about security |
| 01:37 |
|
sfisque |
https://blogs.oracle.com/java/entry/java_ee_7_the_details |
| 01:38 |
|
sfisque |
oh maybe in Servlet 3.1 |
| 01:38 |
|
sfisque |
jsr 340 |
| 01:39 |
|
whartung |
jsr 196 |
| 01:41 |
|
whartung |
I have to run off -- have a good weekend -- game strong. |
| 01:44 |
|
sfisque |
o/ |
| 03:53 |
|
|
MasterProgram joined ##javaee |
| 04:47 |
|
|
MasterProgram joined ##javaee |
| 04:59 |
|
|
MasterProgram joined ##javaee |
| 07:21 |
|
|
apric left ##javaee |
| 07:37 |
|
|
Voyage joined ##javaee |
| 09:13 |
|
|
Yegor joined ##javaee |
| 09:16 |
|
|
ibaca joined ##javaee |
| 09:17 |
|
Yegor |
Hi all. Someone can help me with EJB? I have two beans http://pastebin.com/nWZr7E6f. Why is my @PostConstruct method called twice? |
| 09:45 |
|
|
kotten_ joined ##javaee |
| 10:15 |
|
|
AlexCzar joined ##javaee |
| 11:14 |
|
|
pdurbin joined ##javaee |
| 14:20 |
|
|
firebird1 joined ##javaee |
| 17:01 |
|
|
firebird1 joined ##javaee |
| 18:47 |
|
|
AlexCzar joined ##javaee |
| 20:41 |
|
|
sajjadg joined ##javaee |
| 21:04 |
|
|
Voyage joined ##javaee |
| 21:09 |
|
|
SoniEx2 joined ##javaee |
| 21:16 |
|
|
MasterProgram joined ##javaee |
| 21:21 |
|
|
NationalCoder joined ##javaee |
| 21:49 |
|
|
Voyage joined ##javaee |
| 21:58 |
|
|
vadian joined ##javaee |
| 21:59 |
|
|
vadian left ##javaee |
| 22:24 |
|
|
NationalCoder joined ##javaee |
| 22:30 |
|
|
MasterProgram joined ##javaee |
| 23:02 |
|
|
NationalCoder joined ##javaee |
| 23:18 |
|
|
Voyage joined ##javaee |
| 23:37 |
|
pdurbin |
https://www.openshift.com/blogs/how-to-host-your-java-ee-application-with-auto-scaling |
| 23:44 |
|
Voyage |
Can I use spring on the top of a JavaEE application that has all features e.g EJBs, JPA, or viceversa as make JavaEE on top of spring ? |
