I don't if you noticed it already but i cloned your repro and i started to create research like folder inside it. I tried to figure out so far:
I checked it out. Great work! I am gonna fill some of the questions inJee-Bee wrote: ↑Mon May 09, 2022 9:30 am You can check it out. Ithink it helps to define a more clear vision how and why some things should be done...
My repro is here: https://github.com/Jee-Bee/FreePDM/tree/conceptod
), but imho "role" means "select a role and you are then stuck with a predefined set of attributes", don't mind roles at all, but the point is then that the attributes in a role should not be fixed, but freely configurable...The question is: What do you want? This proposal will probably become a WB in the future. And that is it, nothing more. If you want to use it, that is good but if you don't want to use it, that is also good. It will become a PDM, with lots of features (that we even didn't thought out very well). Take a look at https://github.com/grd/FreePDM
Yes, you can be right, but in the past I also worked wit companies that didn't have a PDM and I can tell you: Never again!"But to think about it, this should also apply to the data source."
yes please, believe that is the right architectural choice.
at the same time, imho, this inherently means that you have to roll your own wrapper around any version scheme used, but you can leverage on this to overcome the limitation of the typical sw-version schemes that in my simple mind are one-dimensional, i.e. they have a revision, whereas a typical current pdm (imho) is two dimensional, it has a revision and a "state" or "status", which basically are independent (suppose it depends on business logic if you are allowed to go back in status or not). I know of some attempts to get the branching/forking, i.e. non-linear versioning type of schemes into enterprise pdm's, personally I think it will just create a mess out of everything, both for users and downstream, but maybe people will like it.
This is about the role that the user can select. Personally I don't mind roles. An architect has different needs than a mechanical engineer. And maybe we are gonna end up with one role, but I like predefined attributes that I can modify. We just need to figure that part out."That is why we need to have roles and an Admin page where you can pick which role you can select."
hm, might be understanding of wording (where mine can be completely off
No. It is a lot more than that.is it not really a "document management system" that is most tuned towards pdm-like?
I meant local as on a user PC because also a server can be local. I described basically everything that is a network source as a server(Except the raspberry Pi just to strech the basic concept). I think there is more needed than just a hard disc within the network. To be honest a NAS is almost a PC nowadays. So If you like i can add the term network PC in the list even if it don't change much on the working
heda wrote: ↑Mon May 09, 2022 8:45 pm at the same time, imho, this inherently means that you have to roll your own wrapper around any version scheme used, but you can leverage on this to overcome the limitation of the typical sw-version schemes that in my simple mind are one-dimensional, i.e. they have a revision, whereas a typical current pdm (imho) is two dimensional, it has a revision and a "state" or "status", which basically are independent (suppose it depends on business logic if you are allowed to go back in status or not). I know of some attempts to get the branching/forking, i.e. non-linear versioning type of schemes into enterprise pdm's, personally I think it will just create a mess out of everything, both for users and downstream, but maybe people will like it.
There are serval different (Open source)version control systems available. The main difficulty is that those are most written for software.
Where software is released as a whole that is not always the case within the mechanical world. see: @ref to story
This has to be tackeled.
Currently the most promissing implementation is the svn version system (a Client-server).
The most wel known Distributed version systems are git and mercurial.
Yes it is true that the versioning system is the most complicated part.grd wrote: But there are also closed source versions. I think that we should also accept those because in a company they might only allow this kind of software.
...
Why does it need to be tackled? You can also allow them all. But in that case we need to hav an admin page in where we can select which one we use. They all work roughly the same, but we need to make all work of course. But we are gonna start with SVN.
I would argue the reverse. A role (or a set of roles) is given to a user. And based on the roles you have you have a certain freedom. For example a engineer role can de the create new part, checkout, set it ready for release etc. but he can't change for example the library parts that is for example part of the power user.
Thanks i try to theck it outheda wrote: ↑Mon May 09, 2022 8:45 pm comparison between webframeworks: https://www.youtube.com/watch?v=3vfum74ggHE
Within databases there are two main types of databases. SQL based databases. Examples of SQL like databases are SQLite, MariaDB, PostgreSQL.
The other is noSQL. An example of this database is MongoDB
Note: Add What the difference is and how such a database is build up
I try to explain my problem a bit(i was not that far with working outgrd wrote: Why do we need such a database? That is the question. We need it because we can't extract the relevant information out of VCS and that is because a FC file attributes are stored somewhere into that file. And you want to be able to search that metadata information. To be honest, I don't care about which kind of software you use (SQL / noSQL), but SQLAlchemy helps a lot when you use SQL data. That is why I think that we better use SQL. Starting with sqlite.
) that there are basically a few types of databases where i think SQL based databases the most well known are. My issue is that there is as far as i can see some difference in complexity. Maybe from user point of view there is not that much different but i think from implementation wise there is some difference. If SQLAlchemy handles both well to me there is no problem. I wrote it the wrong way around.grd wrote: This is never gonna work. When a part is changed you aso need to update the drawing of that part, the assy (or assy's) in where it belongs and the drawings of those too.
It is true that the browser doesn't mess with your focus. That also counts for a independent GUI. And that is what we want. Why? Because IMO it's better to not run FC when you do your operations and that is because FC copies the files to another directory when it runs the program (that I read somewhere in this thread). So you better run this PDM without using FC.user1234 wrote: ↑Tue May 10, 2022 12:05 am Web based does not mean internet based, it means webbrowser based and can absolutely be local. When it is based on a webbrowser FreeCAD can not steal the focus of the screen (example: when you are middle in a taskdialog, you still can work on the PDM) and is is better organizeable on a screen, at least i think so. Also setting up a web frontend later is much harder else when it is an the start the gui, since can do both, local (computer, network, server) or nonlocal.
Yes. Maybe I was a little bit too enthusiastic and maybe we should only use one. And that counts for the SQL too. Maybe we should just stick with SVN and sqlite. And then in the future we can maybe add others.
I think that the SQL part is probably the hardest thing to do, especially since we need to be able to do things such as "Check Out" (this needs to be achieved in SQL, the VCS doesn't handle this kind of things AFAIK).]
Yes it is true that the versioning system is the most complicated part.
It is good to have an alternative, but I hope that SQLAlchemy "just works".I try to explain my problem a bit(i was not that far with working out
The only thing (but that could also be me) What are the alternatives for SQLAlchemy (https://builtin.com/data-science/python-database) SQL
I still don't fully understand your question. But when a drawing has changed then it should have a new version. That is for sure. And it should also look at the right model.I wrote it the wrong way around.
Let's assume i created a assembly and a drawing(in separate files. Of course this change if the drawing is inside the related part / assembly.) . everything is released and there has to be a change for example i described in the notes a type of glue that is not strong enough. Is it valid to change the drawing without releasing the model. So the drawing get release version 2, but the assembly hold release state 1.
The answer could be "It is up to the user(Read: Admin) / Company" to decide if it's aloud or not. but at least should it be possible in that case
what is check out in this context? At least in git there is something called checkout and svn probably has something at least similar too.I think that the SQL part is probably the hardest thing to do, especially since we need to be able to do things such as "Check Out" (this needs to be achieved in SQL, the VCS doesn't handle this kind of things AFAIK).
