State of Lua web dev with LuaJIT

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
5 messages Options
Reply | Threaded
Open this post in threaded view
|

State of Lua web dev with LuaJIT

Lloyd Dube
Hi everyone,

I'm researching Lua web development options using LuaJIT and OpenResty
- or something that scalable.

Most web frameworks seem to have hit EOL. Lapis seems to have
activity, but is it the only currently mature and active Lua web
framework?

What is the prognosis for LuaJIT? Is there current development or is
there risk of abandonment? I'm not looking for flames, just doing my
due diligence before committing to anything.


Kind regards,
~ Lloyd
Reply | Threaded
Open this post in threaded view
|

Re: State of Lua web dev with LuaJIT

Eretnek Hippi Messiás
hi there! :)

luajit is actively maintained these times, but Mike, its author went silent for like a whole year before, but he signed that he had other businesses to focus on...

nobody can predict the future, while based on its unique awesomeness, i guess luajit wont die out, and on the contrary, based on its high entry level (to maintain its sources, not to use it) there is a chance that its real fate depends on Mike...

otherwise the lj fork of openresty is actively maintained, and even if none of those ppl around it are actually Mike, they arent dumb either :) if im right, then Siddhesh works/worked on gcc (at fedora?), according to his blog that i once found :D he started his own fork, moonjit, when Mike disappeared back in time, but now i believe that he moved all of his effort to the fork that is maintained for openresty. this fork is up to date, compatible, and all their customizations can be disabled on compile time, if im right, furthermore it supports ppc bigendian too, that was dropped from luajit (if im not mistaken here :D if i am, then its little endian, but i dont think so...)

not to forget, moonjit aims to become compatible with the current puc lua, it has a dedicated branch for it, but i dunno how far it reached that way, but i think not too much...

there are also two more important forks that im aware of, one is raptorjit, that Luke started with an extreme level of activity on it, with the main goal to make the jit behave on a more predictable way, but he also dropped much architectures and the custom allocator, just in the name of maintainability... later on, i think he had basically other duties, and Max took the torch from him, but i dont see much activity from his behalf there either, probably they achieved what they wanted, that is some router driver for high traffic (if im right :D )

and there is luavela, that is made under the flagship of iponweb, im not really familiar with it, but i can see some activity around it sometimes, and i know about it that it has a bunch of interesting modifications...

also, i saw a bunch of talented ppl around luajit who actually could touch the codes, there are other forks out there, and even php8 will use dynasm from luajit as its new jit backend, so after all, i think theres enough ppl with more or less of the necessary interest time and knowledge to keep luajit alive in the worst case... i also read sometimes random portions of luajit, and actually i believe its far not as horrible as some say, i think i didnt see anything that i couldnt understand, expect some assemblers for architectures that im not familiar with, or such stuffs that i didnt track down to its roots... however there are parts ive never seen so far now, and it is somewhat mentally overwhelming to keep all the important details on the right track (yummy, i actually like such stuffs! :D )

---

about web stuffs, there is kernelsauce/turbo (on github) that has easily digestible sauce codes and an awesome set of abilities, however currently they are about to slowly tracking down some memory issues, that can even be anything like some innocent dead weight made by table allocations, but it can also be a real memory leak as well... it is intended to be embeddable, and therefore it is really small fast and it is self-contained, so even if it has any bugs, it is a project that anybody can fork any time... (ok, not anybody, but whatever :D ) it doesnt have a rushing development on it, but it is fairly complete, ppl use it in production, and they say it is really fast! :) (im yet to try it seriously, but i have prerequisites and a bunch of nontrivial [dev]life-issues at hand to resolve 1st...)

i hope i could help, i wish all the bests to u, and have fun! :)

Sent with ProtonMail Secure Email.
Reply | Threaded
Open this post in threaded view
|

Re: State of Lua web dev with LuaJIT

Robert Burke
In reply to this post by Lloyd Dube
Hello,

Openresty is core to the business of Cloudflare, Kong, and itch.io. Lapis is core to the business of itch.io. In my view, the reason there is not much activity adding new features to these projects is because they work quite well today.

On Wed, Oct 7, 2020, 08:04 Lloyd Dube <[hidden email]> wrote:
Hi everyone,

I'm researching Lua web development options using LuaJIT and OpenResty
- or something that scalable.

Most web frameworks seem to have hit EOL. Lapis seems to have
activity, but is it the only currently mature and active Lua web
framework?

What is the prognosis for LuaJIT? Is there current development or is
there risk of abandonment? I'm not looking for flames, just doing my
due diligence before committing to anything.


Kind regards,
~ Lloyd
Reply | Threaded
Open this post in threaded view
|

Re: State of Lua web dev with LuaJIT

Lloyd Dube
Thank you for your comprehensive responses, everyone :)

Is it safe to conclude that I can rely on OpenResty and LuaJIT (Lua
ver 5.1)as a platform to build and scale online services, with a
reasonable guarantee of future stability?

I've also looked at Go as an alternative, although I feel like for a
small team (just me, for now) Lua could yield higher productivity/
speed of development.


Kind regards,
~ Lloyd

On Wed, Oct 7, 2020 at 11:52 AM Robert Burke <[hidden email]> wrote:

>
> Hello,
>
> Openresty is core to the business of Cloudflare, Kong, and itch.io. Lapis is core to the business of itch.io. In my view, the reason there is not much activity adding new features to these projects is because they work quite well today.
>
> On Wed, Oct 7, 2020, 08:04 Lloyd Dube <[hidden email]> wrote:
>>
>> Hi everyone,
>>
>> I'm researching Lua web development options using LuaJIT and OpenResty
>> - or something that scalable.
>>
>> Most web frameworks seem to have hit EOL. Lapis seems to have
>> activity, but is it the only currently mature and active Lua web
>> framework?
>>
>> What is the prognosis for LuaJIT? Is there current development or is
>> there risk of abandonment? I'm not looking for flames, just doing my
>> due diligence before committing to anything.
>>
>>
>> Kind regards,
>> ~ Lloyd
Reply | Threaded
Open this post in threaded view
|

Re: State of Lua web dev with LuaJIT

Eretnek Hippi Messiás
> Is it safe to conclude

well, nothing is safe, probably they won't go away, but an extra safety measurement is when u can actually fork anything important from ur stack if the worst would come, and thats why turbo and the whole lua universe is really sympathetic to me :)

go is actually kinda neat, but i dont know it well-enough, however i think that ur assumption about taking the lua path is right, while i would still advise u to never save time on researching, planning and making groundwork when u have to make a critical decision, it will most likely yield ur investment :) my needs probably differ from urs, but i do extreme amounts of those fundamental stuffs i listed, and i dont regret none of my choices for my stack, with lua at its hearth :) (btw even if i wanna go for turbo, i strongly believe that i wouldnt cry if i would take the openresty road, and i also saw some other stuffs that made my decisions very hard, so even around lua there are some other nice gems, just i dunno their names from the top of my head)

bests! :)

Sent with ProtonMail Secure Email.