sqlitebrowser: Parse error at ...: Premature end of document.

[amake (Aaron Madlon-Kay)]

This commit seems to have broken building the port for me: [0c5579b2d0a00c0fdb3907aa5facc8f2eea3c2fb/macports-ports]

See attached log

macOS 10.15.5 19F101

Xcode 11.5 11E608c

[kencu (Ken)]

That's a weird one:

:info:build Parse error at /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_macports_release_tarballs_ports_databases_sqlitebrowser/sqlitebrowser/work/sqlitebrowser-sqlitebrowser-f07dafd/src/translations/sqlb_es_ES.ts:4782:11: Premature end of document.
897	:info:build Updating 'translations/sqlb_es_ES.ts'...
898	:info:build     Found 1010 source text(s) (0 new and 1010 already existing)
899	:info:build Parse error at /opt/local/var/macports/build/_opt_local_var_macports_sources_rsync.macports.org_macports_release_tarballs_ports_databases_sqlitebrowser/sqlitebrowser/work/sqlitebrowser-sqlitebrowser-f07dafd/src/translations/sqlb_fr.ts:3602:68: Premature end of document.
900	:info:build Updating 'translations/sqlb_fr.ts'...
901	:info:build     Found 1010 source text(s) (0 new and 1010 already existing)

Some kind of error parsing and installing the translations???

It built fine for me, and obviously for i0ntempest, and on the CI system, and on (most of) the buildbots... <​https://ports.macports.org/port/sqlitebrowser/builds>

Is it possible you have a bad download file?

[amake (Aaron Madlon-Kay)]

Weird. I cleaned and tried it again and it worked. Then again and it failed. So I wonder if it's a concurrency issue.

Repeatedly rebuilding now seems to have a ~1 in 3 chance of failure.

[i0ntempest]

Ive seen this error when i was tinkering the build, but only for once or maybe twice. Will try to see if I can reproduce later.

[i0ntempest]

Reproduced: built 10 times, 2 failed, and errors are from different files: one from translations/sqlb_zh.ts and one from translations/sqlb_de.ts. I suspect this could be a problem in Qt...

[kencu (Ken)]

more likely a problem with parallel building.

[i0ntempest]

Indeed the more cores you have the bigger chance the build is gonna fail. Just tried on a 6C12T system half of the 20 builds failed. Added use_parallel_build no to the portfile and all of 10 the builds succeeded, though it becomes MUCH slower. So is it worth it?

[amake (Aaron Madlon-Kay)]

I see it's GPL-3 so I guess we can't distribute the binary?

Failing to build in parallel is a bug, so it should be reported upstream.

If we can patch it ourselves then that would be a good solution in the meantime to keep everyone happy.

Otherwise I guess I prefer a predictably successful build to a fast one. It's not like this port gets a lot of updates.

[i0ntempest]

K, I'll report this to upstream and disable parallel build for now.

[i0ntempest]

Don't know if I was lucky but using the newest version 3.12.0, 20/20 builds succeeded on a 4C8T system, though 10/20 still failed on a 6C12T system.

[i0ntempest]

In b76f03daca57496fd95bd96ca17d582c75b7ad48/macports-ports (master):

sqlitebrowser: update to 3.12.0

And disable parallel build until upstream gives a fix
Closes: #60626