Support for multiple repositories about gitinspector HOT 20 CLOSED

I now consider this feature fully implemented, please report any problems in a new issue.

from gitinspector.

Hi.

I have considered this feature before and I don't really know the way to go 
when implementing it. As far as I can see, there are two options here:

1. The statistics from the different repositories could be merged to create a 
truly merged report that still looks like it is from just one repository. Even 
if this could be really useful, the problem with this approach is that 
duplicate rows (or rows very similar to each other with just different 
white-spaces) would be counted as unique rows - simply because we can't use git 
to track them as they are in separate repos.

2. We could just create a separate report from each repo and just spit it out 
the same as when running gitinspector multiple times. In the HTML output it 
could just generate separate tabs for each repo and present them on the same 
page.

Then again; we could just introduce a --merge flag in order to switch between 
(1) and (2).

I'm open for ideas and feedback on how to ac actually go about it. This is 
probably something that would be great to include in 0.4.0, so I'm tagging it 
for that release (for now).

/Adam Waldenberg

Original comment by [email protected] on 3 Feb 2014 at 6:33

• Changed title: Support for multiple repositories
• Changed state: Accepted
• Added labels: Milestone-Release0.4.0

from gitinspector.

I have a feeling that by 1 you mean combining repository information at data 
collection stage. Don't you think it can be done when post-processing 
per-repository reports instead, disregarding cross-repository row duplicates 
completely?

Say repo1 generates the following «historical commit information» 
(Author/C/I/D) report section:

Hugo Doom : 52 : 389 : 291
Lisa Hacker : 29 : 302 : 156

And repo2 generates this:

Hugo Doom : 23 : 144 : 913
Lisa Hacker : 73 : 2102 : 378

The corresponding section from the combined report would then be:

Hugo Doom : 52+23 : 389+144 : 291+913
Lisa Hacker : 29+73 : 302+2102 : 156+378

(Turns out I don't know how changes percentage is calculated so I omitted it. 
But if in repo1 changes of an author are calculated as B1/T2 and in repo2 it is 
B2/T2, the combined value should be (B1+B2)/(T1+T2))

Original comment by [email protected] on 4 Feb 2014 at 12:06

from gitinspector.

Your example describes exactly how I imagine (1) would work.

/Adam Waldenberg

Original comment by [email protected] on 8 Feb 2014 at 1:27

from gitinspector.

Original comment by [email protected] on 24 Nov 2014 at 8:03

• Added labels: Milestone-Release0.5.0
• Removed labels: Milestone-Release0.4.0

from gitinspector.

Is there any workaround for this??

Original comment by [email protected] on 14 May 2015 at 9:03

from gitinspector.

Using gitsubtree is a easy workaround to retrieve stats from different 
repositories

Original comment by [email protected] on 14 May 2015 at 12:28

from gitinspector.

That's probably the best option until this is actually properly implemented.

/Adam Waldenberg

Original comment by [email protected] on 15 May 2015 at 8:58

from gitinspector.

I really need this feature too, looking forward to this.

from gitinspector.

This has been postponed for 0.5.0 because the feature needs some refactoring to the code before this can be implemented in a proper and clean way. After 0.4.x development is done, this feature will be the next thing added together with a new responsive HTML output based on Bootstrap.

from gitinspector.

Cool. I am the original reporter, keeping an eye on this.

from gitinspector.

Work on this has slowly begun.

from gitinspector.

Almost done.

Current issues:

• It only prints out the name of the last repository specified.
• "-m" metrics is missing support for this right now.
• Progress output does not specify which repo is currently being processed.
• Does not filter out duplicate repos.

Please test this with filtering and all the other features of gitinspector and check that it behaves as you would expect.

from gitinspector.

The -m flag (and metrics module) is now functional with multiple repositories.

from gitinspector.

The name of all the repositories should now be properly printed to stdout when generating a report.

from gitinspector.

Hey adam-waldenberg,
Can you please provide whole command to use for multiple repositories? I have following command for me.
python /opt/gitinspector/gitinspector.py -HTlr --grading --file-types=sh,yml,yaml,py,conf --format=htmlembedded <repo1_path> <repo2_path> > output.html
How should I use -m option incase of multiple repos ??

from gitinspector.

-m works the same regardless of if you analyze multiple repositories or not.

Your command looks fine. However, please note that analyzing multiple repos only works with 0.5.0 on the master branch at the moment.

from gitinspector.

Thanks Adam.
I was using .0.4. I will try with version .0.5

from gitinspector.

Hi Adam,
Back again :(
I am able to generate consolidated report successfully. However, this report has all the authors from all the repositories, making it difficult to segregate them. Any option for this? means Any way to identify relationship between author and repository.

from gitinspector.

@yogesh9391 Sorry. I completely missed this comment.

No, there is no way to see what came from what repo. The way to do this is to run gitinspector separately on each individual repo and then run a merged report to get a summary.

from gitinspector.