Re: Fwd: GNU MIG

[Joshua Branson]

Protonmail Me <sjoerd.van.leent@pm.me> writes:

> Dear Hurd Developers,
>
> I attempted to send below e-mail to the hurd-maintainters, but I found out 
> that this e-mail is presented on Savannah, although posted in 2011. I hope 
> that I
> am doing this right.
>
> Aside from what I mentioned in the e-mail forwarded, I want to add two other 
> bullets:
>
> *) I'd like to upgrade GNU MIG's IDL language to a more modern syntax, and 
> enhance MIG with asynchronous handling
>
> *) I totally forgot to mention Emacs LISP and PERL as a possible target 
> languages, my bad

This sounds pretty awesome!  I think I read somewhere that GNU MIG has
an unused bit somewhere everytime that it sends messages...You could set
said bit to a 1, to specify that this is a newer version of MIG.  I wish
I know where it was that I read that.

>
> Further I would have a suggestion at changing the name of Mach Interface 
> Generator to MIG Interface Generator, the latter would be useful for any GNU
> (free) software in the need of either vertical or horizontal IPC.

I should also mention that there are some alternatives to MIG:  FLICK.

https://www.gnu.org/software/hurd/open_issues/rpc_stub_generator.html

>
> Please read on,
> Regards,
> Sjoerd
>
> -------- Doorgestuurd bericht -------- 
>
>  Onderwerp:   GNU MIG  
>  Datum:   Fri, 28 Sep 2018 21:10:40 +0200  
>  Van:   Protonmail Me <sjoerd.van.leent@pm.me>  
>  Antwoord-naar:   Protonmail Me <sjoerd.van.leent@pm.me>  
>  Aan:   hurd-maintainers@gnu.org  
>
> Dear GNU Hurd maintainers,
>
> I have over 10 years of Geospatial (GIS) system implementations, and 
> have found some time recently to actively be able to start producing 
> free software. Within this spare time, I'd like to create a GIS 
> framework, different from libraries such as GDAL and different from 
> applications like QGIS or GRASS, finding them inadequate. Also, having 
> experience with a number of proprietary systems, I can safely say that 
> utility organisations using them suffer more than gaining something out 
> of them. For me, reason enough to embark on this train.
>
> However, having seen what is absent and what is necessary, the first 
> thing that came to mind is a decent, reasonably efficient IPC solution. 
> REST (which is fast enough for most things) just doesn't work for the 
> monstrous data GIS systems produce and consume. Within corporate realms, 
> Protobuf is establishing itself slowly, but is not what I am looking 
> for. Then I remembered that Mach uses MIG and I remembered that the GNU 
> project was still attempting to create a microkernel, and has GNU MIG.
>
> On my own computer, I made a quick test whether GNU MIG could be 
> compiled on GNU/Linux, and without to much hassle (linking some 
> platform-independent Mach stuff in), it did. However, I notice that I 
> can not seem to find any design complete documentation either on 
> functional or technical level. The MIG preprocessor (migcom) does 
> segfault easily and there are some other twitchy things (such as calling 
> mig without any arguments just puts one back in the shell without 
> warning or usage line).
>
> I would be happy to embark on a mission to
>
> * Work out how GNU MIG works and write some documentation (if this is 
> wanted, I would like to know whether to use GNU Info or RST and whether 
> some documentation is available somewhere)
>
> * Make GNU MIG a bit more stable
>
> * Extract platform requirements such that GNU MIG can be used on both 
> GNU Hurd and GNU/Linux
>
> * Provide a C++(17?) generator
>
> * Provide other generators (Python, GNU Guile, etc)
>
> Of course, if others are already doing this, I would rather just like to 
> join them
>
> Kind Regards,
> Sjoerd van Leent