Is webrev generation still relevant?

Andrew Dinn adinn at redhat.com
Thu Jan 18 11:50:14 UTC 2024


On 18/01/2024 09:04, Magnus Ihse Bursie wrote:
> At the onset of Project Skara, one goal was to keep backwards 
> compatibility with developers' workflows. For this, a Skara bot was 
> created which generates webrevs, as closely aligned to the original ksh 
> webrev script as possible.
> 
> Now I believe all developers are well into the Skara/GitHub way of doing 
> things, and I have not heard someone refer to webrevs in a long time. So 
> my first question is:
> 
> * Is it still relevant to continue let the Skara bots generate webrevs?
No noise can be a sign that something is functioning smoothly and 
efficiently.

I almost always use webrevs in preference to the github changes. The 
different viewing options, in particular the frames view, give me a much 
better overview of what is being changed.

For me webrevs are essential when a large complex change is made. Just 
to give two recent examples, I would have found it much harder to review 
the recent AArch64 poly1305 changes or the changes to the 
ConstantPoolCache field entries without a webrev. These both involved 
adding multiple changes sets per file and correlating those changes via 
the github UI is very difficult. The frames view allows you to skip back 
and forwards in both new and old code to see how a suite of changes 
relate to each other *and* to the *unchanged* code.

regards,


Andrew Dinn
-----------

Directors: Michael Cunningham, Michael ("Mike") O'Neill



More information about the jdk-dev mailing list