Open vietj opened 3 years ago
Could you provide a bit more info about the scope of the Oracle client item?
we are planning this for 4.2
On Tue, Jul 6, 2021 at 1:36 AM Ángel Sanz @.***> wrote:
Could you provide a bit more info about the scope of the Oracle client item?
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/vert-x3/issues/issues/566#issuecomment-874369186, or unsubscribe https://github.com/notifications/unsubscribe-auth/AABXDCQ5I6XWPFMKHMOKR2TTWI6W3ANCNFSM4UQ7376Q .
Hi Julien, thanks for the prompt reply. My question was about the scope of the feature. Is it a full Oracle DB client which does not use JDBC?
this is actually a wrapper of Oracle JDBC Reactive Extensions (https://docs.oracle.com/en/database/oracle/oracle-database/21/jjdbc/jdbc-reactive-extensions.html#GUID-1C40C43B-3823-4848-8B5A-D2F97A82F79B)
it is not possible to perform or implement wire level access for Oracle for legal or technical reasons.
On Tue, Jul 6, 2021 at 1:20 PM Ángel Sanz @.***> wrote:
Hi Julien, thanks for the prompt reply. My question was about the scope of the feature. Is it a full Oracle DB client which does not use JDBC?
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.
@vietj what would be the best place to add PcapWriteHandler
to the pipeline? Asking because when I added it in the same place as the LoggingHandler
, things did not work becayse PcapWriteHandler
needs to implement the channelActive
method.
it is not clear what you mean ?
I am wondering what the proper place in Vertx is for adding the PcapWriteHandler
to the Netty processing pipeline is.
If I just add it in the same places where LoggingHandler
is being setup, then PcapWriteHandler
results in NPEs because its channelActive
method has not been called (I presume because it was added too late).
that sounds weird, do you have a stack trace ?
I'll have to build everything again and capture it. I'll to do that later on today or tomorrow.
do you have a branch you can share too ?
https://github.com/geoand/vert.x/tree/4.2.2-custom is what I had done.
But in order to get that working, I needed to also patch Netty and implement handlerAdded
in PcapWriteHandler
to simply do what channelActive
does
so the issue is that the handler is added too late and some handlers like logging handler do support it, but not the pcap write handler, is that right ?
Exactly, that is how I understood the issue
did you try subclassing the handler and have the handlerAdded call the channelActive method as workaround ?
I didn't, but I assume that could work.
just to see if that works
nevertheless it would be good to have this handler in netty lazily handle things
Yup, for sure. I'll take another look tomorrow
also this handler is final, so it could work with a wrapper I think
I have asked on https://github.com/netty/netty/issues/10385 to see if it's safe to implement handlerAdeed
as well. I asked because I was seeing some weird behavior which I am not sure is due to me doing something wrong, or handlerAdded
being incompatible with PcapWriteHandler
handlerAdeed
seems to be safe, but I am having problems writing the client and server capture to the same file so I've asked again on the netty issue.
can the file path be specified ?
I think the file name should be specified when creating the client or the server with something like a pattern in options ?
Yes absolutely.
In what I have now, if the server and client files are different, the capture works properly.
But I really wanted to have the capability to use the same file for all capturing. We'll see if it can happen or not.
I think can we use something like a Function<SocketAddress, AsyncFile>
on NetClient
, NetServer
etc... that would be configured with options by default.
so default would be to use options, but user can specify a function for ultimate flexibility
👍🏼
I'll wait and see what the Netty folks say about using a single file and then come back to this.
I found a few more cycles to take another look at this and actually it seems like the only issue is to properly manage writing the data from multiple sources into the output file.
Now I just need to think of what the proper way of doing that would be
https://github.com/eclipse-vertx/vert.x/pull/4260 is a first draft of what I have in mind
I think the http 3 implementation should be prioritized over grpc transportability, that http3 has become the standard now.
QUIC support is a must!
@vietj May I know if io.vertx.core.http.HttpServerRequest can be updated in next release to remove the deprecated class javax.security.cert.X509Certificate for Java 17 compatibility ?
Thanks
@vietj May I know if io.vertx.core.http.HttpServerRequest can be updated in next release to remove the deprecated class javax.security.cert.X509Certificate for Java 17 compatibility ? Thanks
In what way does this block Java 17 for you?
Other stuff
4.x, x > 3
4.4
Future
JsonObject
Single
andMaybe
4.3
Committed
jackson-dataformat-*
as it forces pullingjackson-databind
dependency when the use of the library is purely internalBest effort
4.2
Expected
ROUTING
)Others