x11/xscreensaver: Update to 6.04

[zirias]

Summary

Any code for calling an external helper for PAM was removed in 6.03,
therefore replace pam_helper by unix-selfauth-helper that's called by
pam_exec.so

Also fix line length in USES.

[zirias]

don't hardcode /usr/local in pam configuration

[zirias]

rebase on latest main

[zirias]

Latest upstream version is now 6.04

Update patch accordingly.

[zirias]

Fix icon in xscreensaver-start.desktop

[zirias]

Add run dep on p5-LWP-Protocol-https

The new "mapscroller" hack needs LWP::Protocol::https.

[zirias]

Re-apply accidentally reverted icon fix.

[zirias]

rebase after created-by removal

[zirias]

rebase after gdk-pixbuf change

[cy]

Please submit this as two patches, one dependent on the other.

x11/xscreensaver/Makefile

20

This is a cosmetic change and should be a separate commit.

Changes like in an update "something" mess up history making it difficult for those of us who follow looking to bisect some bug.

Though this change is simple enough it's counter to the make one change per commit paradigm.

If separate commits would break a port (or build in src) -- see my upcoming (not yet in the tree for a week or two) cde commit -- one would commit the jumbo patch with sufficient documentation in the commit log why all the changes are being done at once.

[cy]

Or document the minor cosmetic change. (This change may be minor enough to commit it as is.)

[zirias]

Seriously, I think it would be pretty ridiculous to make one single line break an extra commit. But I'll happily mention it in the commit message of course ;)

So far, this isn't ready to land anyways cause I'm still waiting for an EN fixing a bug in pam_exec for 13.x-RELEASE...

[cy]

In D35348#817241, @felix_palmen-it.de wrote:

Seriously, I think it would be pretty ridiculous to make one single line break an extra commit. But I'll happily mention it in the commit message of course ;)

Your commit log message might read,

Update to 6.0.5.

Any code for calling an external helper for PAM was removed in 6.03,
therefore replace pam_helper by unix-selfauth-helper that's called by
pam_exec.so.

Minor formatting adjustments.

You don't need to articulate each adjustment. You need to mention that adjustments outside of the primary change will be made.

So far, this isn't ready to land anyways cause I'm still waiting for an EN fixing a bug in pam_exec for 13.x-RELEASE...

[zirias]

In D35348#817314, @cy wrote:

You don't need to articulate each adjustment. You need to mention that adjustments outside of the primary change will be made.

That, of course, totally makes sense and I'll keep it in mind in case I'll be able to commit it, we will see. Nothing against separating formatting changes into separate commits of course, I practice that myself, I just think it's a bit over-the-top for one simple added linebreak ;)

[cy]

In D35348#817320, @zirias wrote:

In D35348#817314, @cy wrote:

You don't need to articulate each adjustment. You need to mention that adjustments outside of the primary change will be made.

That, of course, totally makes sense and I'll keep it in mind in case I'll be able to commit it, we will see. Nothing against separating formatting changes into separate commits of course, I practice that myself, I just think it's a bit over-the-top for one simple added linebreak ;)

Back in the CVS days we the guidance regarding ports commits was loose. People tended to patches into commits because CVS was difficult to use. When we switched to SVN the guidance was to batch patches into larger commit packages. The reason for this was that each commit caused significant repo bloat. Now that we use GIT the bloat problem is no more. Additionally, committing to a repo is cheap and pushing multiple commits is easy.

I altered my approach to commits when I got my src commit bit. In src they tended to break down commits into functional and cosmetic changes, making it easier to follow along when reading history. Many ports committers have adopted this same strategy. I've noticed that since we cut from SVN to GIT that more people comment negatively about "jumbo" commits than in the past. It's my impression that ports approach is becoming more like that of src in this regard. This would be my advice.

Now that you are a committer, I trust that your mentor (tcbrener) will guide you through the process. As such, the patch looks good. Work with tcbrener@ to commit it.

When you commit this the commit log should say reviewed by those who have accepted the revision and approved by your mentor. tcbrener@ will guide you through this too.

Have fun. ;)

[zirias]

Rebase onto current main, add note about the cosmetic fix to USES.

[tcberner]

[manu]

Not tested but LGTM.