Feature/physical size - [merged] about zathura

[sebastinas]

On GitLab by @knuesel on Feb 12, 2018, 11:25

---

added 4 commits

• d625c0d - Distinguish zoom and scale, always apply DPI correction on scale
• b4eca29 - Replace monitor "dpi" with "ppi"
• 5505575 - Assume PPI=100 (typical value) if info unavailable
• f6dd368 - Fix compilation for GTK+ < 3.22

Compare with previous version

[sebastinas]

Is this just for the sake of renaming of does the meaning of the value change as well? If it's only the former, please don't. If it's the latter, it requires migration code in both database backends.

[sebastinas]

We can easily rename it everywhere in the code, but changes to the database on-disk format always need to have the code that handles upgrades from previous versions.

[sebastinas]

On GitLab by @knuesel on Feb 12, 2018, 14:46

Commented on zathura/database-plain.c line 27

---

There is a change in meaning:

Currently, zoom is the same as scale: it's the number of pixels per document
point. In particular, zoom 100% means 1 pixel per point.

This MR redefines zoom 100% as "same size on screen as on paper". The scale is
still the number of pixels per point, and this value is now different from the
zoom (it's a function of the zoom and the monitor DPI/PPI).

These meanings for zoom and scale actually match the old code in many places: zoom
was used for the user concept in girara settings, and scale was used wherever a
"pixel per point" factor was needed.

Unfortunately, "scale" was also used as the name to store the zoom in the
database and in data structures (such as mark and link target) where it really
meant the user setting and not the "pixel per point". This is what I propose to
change here.

Regarding the database upgrade: do you think it's worth migrating 'scale' to
'zoom' it if the meaning has changed anyway? Should I just add code to delete
'scale' if present?

[sebastinas]

On GitLab by @knuesel on Feb 12, 2018, 14:50

---

added 1 commit

• 1a12214 - Bump GTK+ to 3.22

Compare with previous version

[sebastinas]

On GitLab by @knuesel on Feb 12, 2018, 15:18

---

added 1 commit

• 4f3837c - Remove Cairo 1.14 version checks

Compare with previous version

[sebastinas]

Okay, in that case it should be enough to upgrade the sqlite based database by dropping the old scale column and adding a new zoom column.

[sebastinas]

On GitLab by @knuesel on Feb 13, 2018, 19:50

Commented on zathura/database-plain.c line 27

---

OK but I wonder if it wouldn't be best to leave the scale column in place, since dropping a column is a bit complex in sqlite, and the scale column wouldn't interfere in any way?

[sebastinas]

Yes, that would suffice. Sample code to add the new column on demand is readily available in the sqlite backend.

[sebastinas]

On GitLab by @knuesel on Feb 13, 2018, 20:01

---

added 1 commit

• 09d8a37 - Add zoom column to sqlite database if missing

Compare with previous version

[sebastinas]

On GitLab by @knuesel on Feb 13, 2018, 20:01

Commented on zathura/database-plain.c line 27

---

Perfect, commit pushed.

[sebastinas]

This change requires updates to all plugins:

utils.c: In function ‘poppler_link_to_zathura_link’:
utils.c:37:19: error: ‘zathura_link_target_t {aka struct zathura_link_target_s}’ has no member named ‘scale’
             target.scale          = poppler_destination->zoom;
                   ^
Makefile:49: recipe for target 'utils.o' failed

(from zathura-pdf-poppler).

Could you changes this to an anonymous union { double scale; double zoom; } and mark scale` as deprecated? Once we merged these changes in zathura, we can then update all the plugins and drop the workaround before the release.

[sebastinas]

hmm, does this function really take a gpointer instead of a GCallback as second-to-last argument? That smells like a bug in glib.

[sebastinas]

On GitLab by @knuesel on Feb 15, 2018, 18:08

Commented on zathura/types.h line 182

---

changed this line in version 6 of the diff

[sebastinas]

On GitLab by @knuesel on Feb 15, 2018, 18:08

---

added 1 commit

• 8b93be9 - Add scale/zoom union in link target

Compare with previous version

[sebastinas]

On GitLab by @knuesel on Feb 15, 2018, 18:09

Commented on zathura/types.h line 182

---

Sorry, I miss that indeed. I've made the changes.

[sebastinas]

On GitLab by @knuesel on Feb 15, 2018, 18:12

Commented on zathura/callbacks.c line 247

---

Yes it also causes a compilation warning (for good reasons)... I'm not sure what should be done about this.

[sebastinas]

Elsewhere we store the signal handler id in the zathura object and use g_signal_handlers_disconnect. Couldn't this work here too?

[sebastinas]

On GitLab by @knuesel on Feb 16, 2018, 16:43

Commented on zathura/callbacks.c line 247

---

changed this line in version 7 of the diff

[sebastinas]

On GitLab by @knuesel on Feb 16, 2018, 16:43

---

added 1 commit

• af90d40 - Store monitors-changed handler for disconnection

Compare with previous version

[sebastinas]

On GitLab by @knuesel on Feb 16, 2018, 16:47

Commented on zathura/callbacks.c line 247

---

Yes, I've pushed changes accordingly.

[sebastinas]

resolved all discussions

[sebastinas]

@knuesel Anything else you want to change before merging this MR? I've played with it a little on a single monitor HiDPI setup and it looked go to me.

[sebastinas]

On GitLab by @knuesel on Feb 19, 2018, 20:07

---

@sebastinas I just tried it on a double-monitor setup today and there are a few issues I'd like to fix:

• a change in monitor resolution is detected, but no action is taken except to store the new pixel-per-inch value. This should be trivial to fix (render_all, refresh_view)
• window moved to another monitor is not detected. A possible fix would be to connect to the top widget's configure-event. The callback would be called repeatedly when a window is being or resized, but the computation should be negligible, e.g. causing refresh only if the PPI value changed). Would you be OK with that?

[sebastinas]

Okay. Yes, that should be fine.

[sebastinas]

On GitLab by @knuesel on Feb 23, 2018, 15:30

---

added 3 commits

• 1e170f0 - Indentation fix
• 8e3f056 - If new PPI is significantly different, yes render and refresh
• 5d362c2 - Listen on configure event for PPI changes

Compare with previous version

[sebastinas]

On GitLab by @knuesel on Feb 23, 2018, 15:36

---

added 1 commit

• 063b234 - Listen on configure event for PPI changes

Compare with previous version

[sebastinas]

On GitLab by @knuesel on Feb 23, 2018, 15:49

---

added 17 commits

• 063b234...dfc31a3 - 16 commits from branch pwmt:develop
• f1ebe9e - Merge remote-tracking branch 'upstream/develop' into feature/physical-size

Compare with previous version

[sebastinas]

On GitLab by @knuesel on Feb 23, 2018, 16:30

---

I've made the discussed changes and merged develop for the meson changes. Playing a bit on dual monitor setups I didn't find any issue so feel free to merge.

[sebastinas]

merged