search

smiths / AIMSS

Assessing the Impact of MDE (Model Driven Engineering) and code generation on the Sustainability of SCS (Scientific Computing Software)
2 stars 2 forks source link

LBM SOP: Version Control Question #127

Closed smiths closed 2 years ago

smiths commented 2 years ago

@peter-michalski how many of the 24 packages use some kind of version control (git or cvs or whatever)? There seems to be a contradiction in the data in the paper. In Table 7 version control is listed under the "less common" heading, which corresponds to 30-63% of the packages, but in Section 3.7 (Maintainability), only 7 packages don't use an issue tracking system. If all of the packages that use issue tracking also use version control, then that means that 17 (about 70%) use version control. Is it the case where some packages use issue tracking, but do not use version control?

peter-michalski commented 2 years ago

Is it the case where some packages use issue tracking, but do not use version control?

I just went through the report and measurement template. While the above statement is correct, there is indeed a problem with the report, as there are 16 packages (66.7%) that appear to use version control.

13 packages on GitHub (all GitHub) appear to use version control 2 packages on GitLab (all GitLab) appear to use version control 1 package uses CVS - This was omitted from paragraph 3 of Section 3.7, and this omission from that text must have led to the propagation of the error into the table.

Fixing:

The relevant part of Section 3.7 should be corrected to: "Of the other packages, one package (SunlightLB) uses CVS for issue tracking and version control. "

In Section 5: Comparison of Artifacts to Research Software in General: The second paragraph needs to be updated to:

All of the top four AHP ranked packages, ESPREsSo, Ludwig, Palabos, and OpenLB, have each of the commonly found artifacts, except only three of them (ESPREsSo, Ludwig, and OpenLB) have a requirements specification or theory notes, and only three of them (ESPREsSo, Ludwig, Palabos) appear to use a version control system.

This appears to be the extent of the damage. FYI the balance of the packages use email or SourceForge for issue tracking.

Let me know if you want me to make these updates.

Edit:

This rests on the assumption that CVS was used for version control in SunlightLB. Otherwise everything is fine. However I believe that it is a safe assumption since CVS stands for Concurrent Versions System, and there is a version (1.1) of the package.

smiths commented 2 years ago

Yes, I think it is a safe assumption that CVS was used for version control.

I've updated the text in the tex file in commit 337087b.

Thank you @peter-michalski.