Re: [HACKERS] Implementing a new Scripting Language

[Igal @ Lucee.org]

On 1/27/2016 8:07 PM, Chapman Flack wrote:

I guess getting some time in playing with PostgreSQL and installing
PL/Java would be the right way to start.  Discussion that's specific
to PL/Java and might not interest all of -hackers can also happen on
pljava-...@lists.pgfoundry.org.


Sounds like a good place to start.

I will continue the conversation at pljava-...@lists.pgfoundry.org


Igal


--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] Implementing a new Scripting Language

[Chapman Flack]

On 01/27/16 22:26, Igal @ Lucee.org wrote:

> If I can help with anything with the pl/Java project I'd love to help.

Man, you do NOT know what you just walked into.  :D

The most imminent thing I see happening is s/1.5.0-SNAPSHOT/1.5.0-BETA1
which is not far off, so testing is always good. Maybe a good opportunity
for you to try it out, get some Lucee code running ... but by separately
compiling and installing jars at first, no JSR two-twenty-three for the
near future. See if it even does what you're hoping it'll do. Depending
on what's reported during beta, there could be delightful diagnostic
diversions too.

I guess getting some time in playing with PostgreSQL and installing
PL/Java would be the right way to start.  Discussion that's specific
to PL/Java and might not interest all of -hackers can also happen on
pljava-...@lists.pgfoundry.org.

Cheers,
-Chap


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] Implementing a new Scripting Language

[Igal @ Lucee.org]

On 1/27/2016 7:12 PM, Chapman Flack wrote:


Now that you mention it, it isn't officially in a ticket. Though it's
not like I was going to forget. :)  I can guarantee it won't be in 1.5...

Speaking of tickets, I should probably make actual tickets, for after
1.5.0, out of all the items now showing in the "known issues" section
of the draft release notes. More chores


Well, I'm not very familiar with PostgreSQL yet, but am trying to learn.

If I can help with anything with the pl/Java project I'd love to help.


Igal


--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] Implementing a new Scripting Language

[Chapman Flack]

On 01/27/16 16:31, Igal @ Lucee.org wrote:

> This can be a major thing.  I will open a ticket in
> https://github.com/tada/pljava -- or is it already on the roadmap?

Now that you mention it, it isn't officially in a ticket. Though it's
not like I was going to forget. :)  I can guarantee it won't be in 1.5...

Speaking of tickets, I should probably make actual tickets, for after
1.5.0, out of all the items now showing in the "known issues" section
of the draft release notes. More chores

-Chap


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] Implementing a new Scripting Language

[Igal @ Lucee.org]

On 1/27/2016 11:47 AM, Chapman Flack wrote:


Thanks. :)  On occasions in the past I have written it
correctly ... there is evidence in the archives 
I believe that!  I actually never remember if it's 223 or 233 and I 
always google it before I post, so in a way I cheated ;)



we will be able to use Javascript
through that via the Nashorn project,

Yes, that's an attraction for me, as the JSR-two-twenty-three
engine for JS is now typically already included with the JRE.
(Technically, I keep reading that there is no two-twenty-three
engine that *needs* to be included and it's totally up to the
JRE packager ... but there have actually been some JavaScript
snippets in PL/Java's build process since July 2014 and no one
has yet reported they broke the build.)
This can be a major thing.  I will open a ticket in 
https://github.com/tada/pljava -- or is it already on the roadmap?



--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] Implementing a new Scripting Language

[Chapman Flack]

On 01/27/2016 02:06 PM, Igal @ Lucee.org wrote:

> two-twenty-three, not thirty ;)),

Thanks. :)  On occasions in the past I have written it
correctly ... there is evidence in the archives 

> we will be able to use Javascript
> through that via the Nashorn project,

Yes, that's an attraction for me, as the JSR-two-twenty-three
engine for JS is now typically already included with the JRE.
(Technically, I keep reading that there is no two-twenty-three
engine that *needs* to be included and it's totally up to the
JRE packager ... but there have actually been some JavaScript
snippets in PL/Java's build process since July 2014 and no one
has yet reported they broke the build.)

-Chap


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] Implementing a new Scripting Language

[Igal @ Lucee.org]

On 1/27/2016 10:48 AM, Chapman Flack wrote:


Ok, if your 233 support is already in beta, you'll get there
before we do, but the paths should intersect eventually. :)


Actually, once your support for JSR-223 is implemented (it's 
two-twenty-three, not thirty ;)), we will be able to use Javascript 
through that via the Nashorn project, which seems like a better (at 
least as a cross-platform) solution than V8 or Chakra.

http://www.oracle.com/technetwork/articles/java/jf14-nashorn-2126515.html

Now that would be very exciting IMO!


Igal




--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] Implementing a new Scripting Language

[Chapman Flack]

On 01/27/2016 01:17 PM, Igal @ Lucee.org wrote:

> the next version of Lucee (currently in Beta) does support JSR-223,
> which I actually mentioned as a viable solution in my first email in

Sorry, I jumped in late.

> this thread.  That would be awesome if PL/Java would support JSR-223.

Ok, if your 233 support is already in beta, you'll get there
before we do, but the paths should intersect eventually. :)

-Chap


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] Implementing a new Scripting Language

[Igal @ Lucee.org]

On 1/27/2016 10:41 AM, Pavel Stehule wrote:


2016-01-27 19:37 GMT+01:00 Pavel Stehule >:



David Fetter wrote some presentation - some years ago was popular
to write own PL

me too

https://wiki.postgresql.org/images/a/a2/Plpgsql_internals.pdf

source code of plpgsql is good example - it pretty simple

you have to write handler

https://github.com/petere/plsh


same author

https://github.com/petere/plhaskell
https://github.com/petere/plxslt


Thanks!  I'll take a look.


Igal

Re: [HACKERS] Implementing a new Scripting Language

[Pavel Stehule]

2016-01-27 19:37 GMT+01:00 Pavel Stehule :

>
>
> 2016-01-27 19:21 GMT+01:00 Igal @ Lucee.org :
>
>> On 1/27/2016 9:57 AM, Vladimir Sitnikov wrote:
>>
>>> That is a good question. ChakraCore has been open sourced recently. It
>>> might be easier to build under Windows.
>>>
>> interesting.  but now we will need to write an extension for that, e.g.
>> PL/Chakra, which brings back my original question:
>> are there any docs as to how to implement a new scripting language? ;)
>>
>
> David Fetter wrote some presentation - some years ago was popular to write
> own PL
>
> me too
>
>  https://wiki.postgresql.org/images/a/a2/Plpgsql_internals.pdf
>
> source code of plpgsql is good example - it pretty simple
>
> you have to write handler
>
> https://github.com/petere/plsh
>

same author

 https://github.com/petere/plhaskell
https://github.com/petere/plxslt

>
>
>
>
> I am not sure you would be able to bind high performance java runtime
>>> with the backend. There are no that many JREs, and not much of them
>>> are good at "in-backend" operation.
>>>
>>> Thus your question boils down to 2 possibilities:
>>> 1) You execute Lucee in some JRE that runs in the backend (frankly
>>> speaking, I doubt it is a good way to go)
>>>
>> yes, that's what I had in mind.  I wasn't thinking of an embedded JRE.
>> TBH I didn't think there were any of those until your email.
>>
>> thanks,
>>
>>
>> Igal
>>
>>
>>
>> --
>> Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
>> To make changes to your subscription:
>> http://www.postgresql.org/mailpref/pgsql-hackers
>>
>
>

Re: [HACKERS] Implementing a new Scripting Language

[Pavel Stehule]

2016-01-27 19:21 GMT+01:00 Igal @ Lucee.org :

> On 1/27/2016 9:57 AM, Vladimir Sitnikov wrote:
>
>> That is a good question. ChakraCore has been open sourced recently. It
>> might be easier to build under Windows.
>>
> interesting.  but now we will need to write an extension for that, e.g.
> PL/Chakra, which brings back my original question:
> are there any docs as to how to implement a new scripting language? ;)
>

David Fetter wrote some presentation - some years ago was popular to write
own PL

me too

 https://wiki.postgresql.org/images/a/a2/Plpgsql_internals.pdf

source code of plpgsql is good example - it pretty simple

you have to write handler

https://github.com/petere/plsh



I am not sure you would be able to bind high performance java runtime
>> with the backend. There are no that many JREs, and not much of them
>> are good at "in-backend" operation.
>>
>> Thus your question boils down to 2 possibilities:
>> 1) You execute Lucee in some JRE that runs in the backend (frankly
>> speaking, I doubt it is a good way to go)
>>
> yes, that's what I had in mind.  I wasn't thinking of an embedded JRE.
> TBH I didn't think there were any of those until your email.
>
> thanks,
>
>
> Igal
>
>
>
> --
> Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
> To make changes to your subscription:
> http://www.postgresql.org/mailpref/pgsql-hackers
>

Re: [HACKERS] Implementing a new Scripting Language

[Igal @ Lucee.org]

On 1/27/2016 9:57 AM, Vladimir Sitnikov wrote:

That is a good question. ChakraCore has been open sourced recently. It
might be easier to build under Windows.
interesting.  but now we will need to write an extension for that, e.g. 
PL/Chakra, which brings back my original question:

are there any docs as to how to implement a new scripting language? ;)

I am not sure you would be able to bind high performance java runtime
with the backend. There are no that many JREs, and not much of them
are good at "in-backend" operation.

Thus your question boils down to 2 possibilities:
1) You execute Lucee in some JRE that runs in the backend (frankly
speaking, I doubt it is a good way to go)
yes, that's what I had in mind.  I wasn't thinking of an embedded JRE.  
TBH I didn't think there were any of those until your email.


thanks,


Igal


--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] Implementing a new Scripting Language

[Igal @ Lucee.org]

On 1/27/2016 9:58 AM, jflack wrote:

I just did a quick search on Lucee and what I found suggests that
it compiles to JVM bytecode and runs on a JVM. If that is the
case, and it can compile methods that will have the sort of
method signatures PL/Java expects, and you can put the .class
files in a jar and load it, your job should be just about done. :)

yes, Lucee uses ASM4 to construct class files which are mostly POJOs

Or, you might end up writing thin wrappers in Java, probably
nothing more.

Another possibility: Java has pluggable script engine support
(java specification request 233, see the javax.script package).
Does Lucee have an existing JSR 233 engine implementation?
the next version of Lucee (currently in Beta) does support JSR-223, 
which I actually mentioned as a viable solution in my first email in 
this thread.  That would be awesome if PL/Java would support JSR-223.


thanks,


Igal


--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] Implementing a new Scripting Language

[Chapman Flack]

On 01/27/2016 11:46 AM, Igal @ Lucee.org wrote:

>> Have you checked PL/Java?
> That seems like a good place to start, thanks.  Are there also any docs
> about the subject?

I just did a quick search on Lucee and what I found suggests that
it compiles to JVM bytecode and runs on a JVM. If that is the
case, and it can compile methods that will have the sort of
method signatures PL/Java expects, and you can put the .class
files in a jar and load it, your job should be just about done. :)

Or, you might end up writing thin wrappers in Java, probably
nothing more.

Another possibility: Java has pluggable script engine support
(java specification request 233, see the javax.script package).
Does Lucee have an existing JSR 233 engine implementation?

PL/Java does not _currently_ have JSR233 support, but it is
definitely being thought about ... the idea being, put a Lucee
JSR233 engine jar on the classpath, define it as a new PostgreSQL
language (PL/Java will handle the interfacing), and then actually
write stuff like

DO $$echo("Hello World");$$ LANGUAGE lucee;

As I said, in current PL/Java version, JSR 233 is still
science fiction ... but it's "hard science fiction", not the
fantasy stuff.

http://tada.github.io/pljava/

-Chap


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] Implementing a new Scripting Language

[Vladimir Sitnikov]

> If the pl/v8 was easily ported to Windows then I probably wouldn't even try 
> to add Lucee,

That is a good question. ChakraCore has been open sourced recently. It
might be easier to build under Windows.

>That seems like a good place to start, thanks
I am not sure you would be able to bind high performance java runtime
with the backend. There are no that many JREs, and not much of them
are good at "in-backend" operation.

Thus your question boils down to 2 possibilities:
1) You execute Lucee in some JRE that runs in the backend (frankly
speaking, I doubt it is a good way to go)
2) You implement Lucee parser/executor/compiler in C and use it as
typical PostgreSQL extension

Vladimir


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] Implementing a new Scripting Language

[Igal @ Lucee.org]

On 1/27/2016 8:40 AM, Vladimir Sitnikov wrote:

Why do you want that at the database level?
Do you have end-to-end scenario that benefits from using Lucee?
Lucee is very intuitive and powerful, so it's more for ease of use than 
anything, and to attract more Lucee users to use PostgreSQL (most of 
them use SQL Server or MySQL).


If the pl/v8 was easily ported to Windows then I probably wouldn't even 
try to add Lucee, but it seems to be quite difficult to compile the 
latest versions for Windows and it looks like the project is not 
maintained (for example, it uses an older version of V8 with no known 
intention to upgrade).


If I had more experience with C++ then I'd probably try to update V8, 
but so far my attempts have not been very fruitful.



I was wondering how difficult it would be to implement a Postgres extension 
that will act as a wrapper around it and will allow to write functions in that 
language?

Have you checked PL/Java?
That seems like a good place to start, thanks.  Are there also any docs 
about the subject?



Igal


--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers

Re: [HACKERS] Implementing a new Scripting Language

[Vladimir Sitnikov]

Why do you want that at the database level?
Do you have end-to-end scenario that benefits from using Lucee?

>I was wondering how difficult it would be to implement a Postgres extension 
>that will act as a wrapper around it and will allow to write functions in that 
>language?

Have you checked PL/Java?

Vladimir


-- 
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers