git - How can DVCS help scientific programming? -


I am doing some initial work in the investigation that how DVCs (guit, hg, market choice) can help in the process For scientific programming, especially for undergraduate students, I feel that I am in a good position for this because I have been programming for some years and am currently starting a postgraduate program in natural science. The goal is to give a brief presentation on it in a month or two.

As far as I see it, apart from the obvious benefits of source control, DVCS currently fulfills the following improvements in a graduate's daily life:

  1. Branches:

    This is bigger than seeing DVCS practices, it is clear that cheap branches primarily encourage the use of new features. Scientific programming is about all experiments. Different branches can be made to zoom parameters or algorithms. This is particularly important because most of the scientific codes have not been able to see even one repeater throughout the lifetime (most graduate students will not even know what is), so the ability to go to different branches will be able to Specific madness will bring in. Comments of Fast Commets can also be used as a rental for lab notebooks.

  2. Pushing the servers:

    Since most of the scientific codes are nowadays run by some kind, cluster, DVCs will be able to use some kind of advanced ADsink Can be used in the form, which are already using HPC groups to push the "production" code

  3. Document collaboration:

    Do I say more? Papers with many authors run like small open source projects. Collaborating on the papers should be a natural fit, when writers write all in latex, with additional complications, if something is written in something like Word, this is where comment comments can play a possible role.

My question is, do you think that DICs can contribute to scientific programmers? I look at many negotiations to go into source control in the community, but most are still looking into subversion. From my cursory notes it seems that for the new graduate students, the DVCs should be the right workflow pattern. Is my thinking wrong? Or is there a scientific coding in the very short time behind heard of DCCS equipment?

  • training One real concern is that I know a lot of particle physicists (big science with large programming projects) whose source is total knowledge of control, the simplified version of cvs checkout , cvs update To run, and CVS commit .

    Yes, CVS I know the leader of a software group who has stepped into Subversion due to these people.

    At the next level of skill, they also know how to specify diff and stat order and branches or tags, but to create or merge branches Can avoid doing so.

    If you are planning to start a DVC, then plan for a deep, continuous training and support program. Scientists (or at least physicists) usually have very little formal training in computer science, and may only be a strange concept of software process.


    Comments

    Popular posts from this blog

    python - Overriding the save method in Django ModelForm -

    html - CSS autoheight, but fit content to height of div -

    qt - How to prevent QAudioInput from automatically boosting the master volume to 100%? -