Hacking:Technical decisions

This page describes technical decisions that stand behind Seamly2D. It will be useful for those who want to know why and which tools we have been using.

The Seamly2D is a standalone project. The reason for such a decision was lack of options. None of projects satisfy our needs: GUI, parametric patterns, support formulas, a pattern format and cross-platform solution. Instead we concentrated on low level blocks.

Historical background

Very often the reason of using one tool or another came before actual start developing a project. It based on founders experience and likes.

• First prototype was made on Pascal.
• Next try was made on C and GTK+. It was time when we describe a pattern in C, instead of using normal interpreter language like Python for example, and export a pattern to png, pdf and ps.
• That prototype then was ported to C++ and GTKMM.
• Next logical conclusion was to separate the data and program. This is where we came up with the idea of a pattern file format based on XML.
• Next thing that you usually want is to have a nice GUI, and here is the time for Qt to show up.

C++ and Qt

When you decide to use Qt you also want to decide which language to use. There are several bindings, but we decided to use a native one - C++. The C++ language itself has a lot of problems and some criticize it a lot. But still it is good language if you need speed and your application is very complex, (many CAD systems are written in C++.) Of course, as was mentioned before, Qt itself is also a big reason why we use C++.

Qt's killer feature for us was Graphics View Framework. This set of classes allow us to quick build the GUI for manipulating objects. A lot of developer teams do their own research in this area, but we decided to concentrate on more important things and not reinvent a wheel. Of course this is not all. Qt is a standalone cross-platform framework with very good documentation and a lot of examples. It brings almost all classes and abstractions we need to build Seamly2D. Almost all because, additionally, we adopted muParser library for our formula parsing.

In the end it allows us to build a project that has only one external dependence: Qt itself. And this fact makes maintaining very easy, which some users appreciate very much.

Code Repository Hosting

The Valentina project was originally hosted on Bitbucket using Mercurial. After the fork which created Seamly2D as a separate development path, the main Seamly2D repository was transferred to Github.

From the beginning was done using Mercurial. In the words of the original programmer: Why? No big reasons. It's more matter of tastes. It just work, it cross platform, it has very nice GUI TortoiseHg. And because of mercurial our main repository lived on Bitbucket. u

Continuous Integration

A process called Continuous Integration (CI) has evolved in recent years. When the Valentina project was hosted on Mercurial, several tools were implemented to support the CI method of releasing code changes and getting those changes into the hands of users quickly. The tools used by the Valentina project for testing software, producing executable packages, and delivering that software are described here .

The Seamly2D project split from the Valentina project by the end of 2017. The Seamly2D code has been moved to a separate repository at https://github.com/FashionFreedom/Seamly2D. The initial Seamly2D CI environment was based on the delivery pipeline used by the Valentina project.

Incorporating minor software changes was still possible until changes happened to the environment provided by Appveyor in the summer of 2018. A change in the default version of a software tool used to build the windows executable caused the automatic process of builds to fail (for windows executables only). Changes still could be and were being made to the Mac executable and to the Ubuntu version which is hosted at the Seamly2D repository on launchpad.

Additional changes in the continuous integration tools provided by Github will force the Seamly2D project to update the pipeline process before January 31, 2019. Please check back to this wiki for updates in the process.