Monthly update April, 2016

[Darius Blaszyk]

Hi all,

I've posted the monthly update for OpSim on the website:
http://opsim.sourceforge.net/. Please share your thoughts and comments
either here or on the website!

Regards, Darius

[Daniel]

Hi Darius,

What is the motivation behind bringing OpSim back from the dead? Are you alone or are you having help from the original developers? Do you have a roadmap?

Regards,

Daniel 

[Hazem]

While I applaud  Darius' efforts and work. I regret to inform you that I am completely out and this is not likely to change anytime soon.  I have been unemployed for over a year and trying to market some software myself.

The mistake I made regarding OpSimwas I decided to run the group as a democracy.  democracy decided to use Firebird as a database, which was in beta, Lazarus was also beta, even the google group, at that time was beta.  They also decided on Python, which is more academic.  They even rejected the name FreeSim, which sounded like a winner.

On a final note, if I may be a little selfish here, I am struggling in understanding "Drag and Drop".  I looked at the code in DWSIM and it seems way too complicated. I am currently using a Visio interface until I figure out "Drag and Drop".  If anyone can point me to a detailed explanation, I will greatly appreciate it. 

---

Date: Mon, 2 May 2016 10:29:28 -0700
From: dani...@gmail.com
To: op...@googlegroups.com
CC: opsim-de...@lists.sourceforge.net; opsim...@lists.sourceforge.net; opsim...@lists.sourceforge.net
Subject: Re: Monthly update April, 2016

--
You received this message because you are subscribed to the Google Groups "Open source Process Simulator" group.
To unsubscribe from this group and stop receiving emails from it, send an email to opsim+un...@googlegroups.com.
To post to this group, send email to op...@googlegroups.com.
Visit this group at https://groups.google.com/group/opsim.
For more options, visit https://groups.google.com/d/optout.

[Darius Blaszyk]

Hi Daniel,
I have been a long follower of OpSim since the day the project was concieved. Wat appealed to me was the fact that it was written in Pascal. I have been developing a long time in Pascal (together with C, Basic, Python, etc.) and found that the FPC RTL and language is one of the best (and most underestimated) out there. I have worked a long time as modeling engineer and done quite some work on interfacing the modeling environment with external tools. For me working on OpSim is pure for fun. I have contact with Samuel, but no other developers. For now I do not mind that I'm the sole developer on OpSim as I need to make some undemocratic decisions. Once the basis is settled, more help would surely be appreciated.
The roadmap for the coming period was posted on the OpSim site jus now (monthly update April). The long term roadmap is getting a basic working application out.
Regard, Darius

[Daniel]

Hi Darius,

That's nice. I understand you, because that's how DWSIM was born and is mantained, purely for fun. :-)

I wish you all the best with this project and, if you need my help or want to do some interfacing with DWSIM, or perhaps use its thermo library, feel free to get in touch. I'm already following opsim updates and will subscribe to the mailing lists.

Best Regards,

Daniel

[Darius Blaszyk]

Hi Hazem,

First of all, too bad you are unemployed now, hope you can soon find the job you are looking for. As for your comments. I can understand some of your frustrations. It's always difficult to get things done your way when working with a (distributed) group like an open source group. I share your frustration regarding that. I'm lucky now in a sense that no active developers are working on OpSim anymore and I can pretty much decide on my own, although I still discuss features and ideas with Samuel, which is the last left of the originators.

I have now decided to split OpSim up into libraries (such as unit conversion, chemical formula and soon to merge into trunk physical properties). This way we can have multiple maintainers working on portions of the code without much interference on the overall code base. Regarding some decisions for OpSim like the use of Firebase I have to fully agree with you. I also think that this was a rather unhappy decision and I am planning to revert this. However not because it's beta software as you describe (which it's not afaik) but rather because of the unneeded dependencies that were added to the application and complexity in the usage. It will take longer but in the end OpSim will benefit more with it's own file format. I'm working on the code and an article describing the rationale behind it. If time permits this will be merged to trunk and posted on the website within a few months. BTW: most of the current code in trunk is planned to be refactored and rewritten in favor of a more modular approach.

I do hope to share some thoughts with you soon about your ideas on the modular way of solving model equations. You wrote an article some time ago and posted it here. I found it very interesting and would like to discuss this with you more in detail. Would this interfere with your other work? If not then please let me know. I already have a basic working equation solver in branches but this is very basic indeed. Always appreciate help with some good ideas!

Regards, Darius

[Matt Henley]

Hi Darius,

I watch for your commit notices and am excited that someone has taken the time to pick up the code and try to do something useful with it. I agree with Hazem that the experience in working on a project that failed was frustrating and painful.  

Glad to hear your thoughts on firebird.  I understand the reasoning behind the decision to use it, but looking back, I think that it was a big mistake.  We spent a lot of time helping people get it working with zeoslib for very little return.. Its overkill for what we were trying to do and a set of dependencies that even now are a pain.

Do you have any links to articles explaining the situation between classes and records with respect to resources and speed?  I do very little programming these days but I didn't know there was a substantial difference in speed.

matt

--

[Darius Blaszyk]

Hi Matt,

Thanks for your message. I can't really comment on what happened in the past as I was not involved. Sorry for the frustrations, but I have been there myself. It's never easy working in teams, let alone in such volatile environments like open source. To maintain quality in OpSim development and keep people involved I find this to be one of my main tasks to concentrate on after I firm up the foundation of the code. In my opinion a more structured approach, demonstrated in my work towards modularizing OpSim, and posting articles on progress (and later on decent proposals and reviews regarding changes of the code), will lead to a better alignment of the team (once / if there will be volunteers to join :) ).

Regarding your question on classes vs records, please check my WIP branch on sourceforge: https://sourceforge.net/p/opsim/code/HEAD/tree/branches/opsim_darius/classes/

There's a readme in that folder with some test results. The overall conclusions are pasted in below in this e-mail You can check the test out from SVN and try for yourself if you like. I was not planning on writing an article on this, unless there will be requests.

Regards, Darius

The results from these tests indicate that records take up about 30-45% less 
RAM. Creating records is 1,5 to 6 times faster and assigning values is 2.3 to 4 
times faster. The creation time is less critical as this is typically done once 
(when loading a file) for large amounts and the overall time is relatively short 
for both methods. However assigning values is done regularly by the solver and 
in great amounts and can amount to a noticeable or even annoying lag of the UI. 
Also the amount of memory used is important as large models can lead to memory 
paging which lead to a significant drop in performance.

[Matt Henley]

Thanks Darius,

I downloaded from SVN and compiled over the weekend.  So far I haven't been able to get Firebird running (the ebuild in gentoo doesn't like the classic version) but everything compiles ok.  No need to write an article.. i was just curious.  I am following the development and applaud your efforts. 

--