Re: We need to do something about build times

Hello.

All the packages I have had poudriere make and hold were ready this morning (TZ=Asia/Tokyo :), 5 hours ago).
However, this equilibrium was broken by glib update 4 hours ago :)
232 ports are now queued again.

If only graphics/cairo is built manually first here, only 2 packages, glib and cairo, will be rebuilt.
However, this eliminates the problem of packages that have lost their cairo, so the next number of packages that need to be rebuilt is reduced to 224.

Continue to rebuild net/avahi-app.
dbus-glib and gobject-introspection, which have lost glib, will also be included in the rebuild.
Only 3 packages were rebuilt, but the number of packages needing to be built next was reduced to 216.

Here print/harfbuzz begins to rebuild.
Rebuilding harfbuzz should be separated from harfbuzz-icu or we may lose a lot of packages that do not need to be removed.
Only 1 was rebuilt, but the rest were reduced to 213.

print/harfbuzz-icu is then rebuilt.
This time the remaining was 212 as usual.

Then, security/p11-kit is rebuilt.
With just this 1 rebuild, the remaining is down to 197.
Also, the reduced number this time includes those related to texlive. Rebuilding of texlive-texmf has been avoided :)

Next, graphics/graphviz is rebuilt.
pango, which lost glib, is also queued at the same time.
Two rebuilds reduced the remaining to 193.
The reduced number this time includes doxygen.

There is no point in continuing to write more :) , but it is likely that running the poudriere manually one by one in this way has reduced the CPU runtime by about 1 hour.
Methods exist to save time, but these methods require human hands.

Regards.