As a system developer from a total Microsoft & Oracle background (for over 10 years) I was converted to open source and I have now changed my direction completely.

One of my first clients needed an erp system so I did some research and played with some oserp systems – My findings – TinyERP , Open Bravo and Compiere made my shortlist and when I started trying to install, configure and customise them I personally found TinyERP was so much better – mainly because simple examples were readily available and it just seemed to work well from the first step onward.

Currently, I am putting a highly customised business solution in place using TinyERP as the basic framework. It really is far better than I expected – I have had some pain but customised development speed is very quick and robust(even as a novice python programmer). I still have some issues with advanced custom reports (I am using OpenRPT for some more complex ones) and I am sticking triggers in the postgresql database for somethings I cannot get TinyERP to do but so far I consider it an excellent product and will be recommending it again.

First of all, thank you very much for taking consideration about TinyERP. I Think the happier should be the guy(s) behind TinyERP actually, but it’s actually good news for anyone claiming to use it.

Small correction again: TinyERP has ALWAYS (at least since 2005) been GPL open source, with source code included in the download.
The only thing new they did last week (and that was said to be planned) was opening their SCM (SVN) to the public.

Still, this was very needed. Indeed, it’s really important when evaluating oss software to be able to track over the time how much the code is changing:
* are the author able to sustain large refactoring efforts or are they only maintaining legacy code?
* is the project dead? (BTW unless is missed something I didn’t see any commit on Neogia trunk since one month)
* how is the code changing? Does it seem to receive a lot of nasty patches to unstable features or are the changes reflecting features increases?
* what kind of code correction was needed to what kind of bug?

And finally, having an open SCM makes receiving community patches and enhancements a lot easier.

One more correction, when you speak about your “activity trends”, all I wanted to tell is that Tiny is receiving half of OpenBravo and five times Compiere forum activity, which isn’t small and isn’t zero as you put in your graph. Also I also tried AlexaRank myself but concluded it was not usable because a Sourceforge hosted project would actually take the AlexaRank of Sourcefore itself. I rather went for number of Delicious bookmarks instead to reflect community interest: http://del.icio.us/url/bf933e46fc974c81a5d3088d4036eae4

OK, finally, I’ve nothing against VC (fund raising) with open source. For instance I’m all the more happy Alfresco exists. And I even think ERP needs some money to fuel it properly because not all ERP code is either funny nor easy to develop or maintain. I’m even happy OpenBravo raised funds and that’s why we decided to partner with them, but that didn’t blind me.

I do believe indeed that cutting edge architecture often does matter a lot more than funding. If VC were to fuel hundreds of thousand of pl/SQL lines of code which makes the code as it unmaintainable nor evolve-able. Then what, they might rewrite all? They could but it’s more like they will be starting from scratch with money but unproven technical skills (yet). That might success eventually but it’s much more likely the cutting edge projects will rather take the lead, finally raising funds eventually.

And on the contrary I saw so many well funded non cutting edge project dying slowly later on, that’s why I believe a lot more in project minimizing the number of line of code per feature so that they can be ported form technological platforms to newer as technological revolutions arise (TinyERP could be ported to Jython and have lower layer externalized to Java code for instance).

And this isn’t only about technology. By design, in TinyERP, verticalizing the ERP to account for instance for perishable products is only about overriding the product object and a few of its methods (a few lines of code, see existing modules). On the contrary, in Compiere or OpenBravo, if it’s easy to add a limit date in the product structure, it’s not easy at all to make suddenly all existing stock valorization or inventory picking methods take into account that new field. You would even barely reuse them, forcing you to spend time re-implementing and maintaining the duplicated logic. It’s the whole difference between real OOP all the way and basic OOP backed with procedural programming. I think it does finally matter for you target audience because it will mostly determine how much verticalizations will exist for free (should be cheap to build), finally bringing an ERP that will match more closely the needs of the end user. Same remark goes for embedding a true BPM engine in TinyERP.

Again don’t get me wrong. I’m happy OfBiz and OpenBravo exist, I wish them luck and we will integrate them with the IT firm (Smile) I work eventually (we even partner with OpenBravo). Still not everything is always black or white and I do have some criticism about OpenBravo. But I would say, those aren’t absolute criticisms, they are only relative to what I think is better. I didn’t wrote them for free bashing but rather because I felt like TinyERP was was somewhat unfairly discriminated. Finally, once against being open to criticism and having to face them is what makes open source better than proprietary software. So OpenBravo and Compiere might very well tackle the criticism about having tire table references from the former Goodyear dev in their code and clean their code. Everyone would win and no one could bash them anymore. May be just like my bashing about TinyERP made them open their SCM a bit faster.

OK, Finally I’ll release a 60 pages white paper about oss ERP myself in some two weeks. Unfortunately, because I was forced to, it’s in French. But fortunately Google translations tools makes that at least readable so I’ll let you know. Regards,

Raphaël Valyi.