In preparation of upcoming work on the serialization, this commit adds a new
testng option to the Bio-Formats automated tests allowing to define a directory
for saving cache (memo) files. If configured, this cache directory can be
later reused via a new dedicated FormatReaderTest.

…iguration

Also adds testMemoFileUsage to a new "memoizer" group so that it can be
run separately if needed.

If testng.cacheDirectory is set, then both the current and cached memo
files will be tested.  Otherwise, only the current memo file is tested.

See discussion on https://trello.com/c/x2nmObSk/191-bfmemo-change-detection

gen-file-list requires testng.directory, testng.configDirectory, and
testng.file-list to be set; the latter is the path to which the list
will be written.

See https://trello.com/c/x2nmObSk/191-bfmemo-change-detection

This allows test-automated to use a file listing created by
gen-file-list, instead of scanning a directory.  Using test-automated in
this way now looks like:

ant -Dtestng.configDirectory=/path/to/config -Dtestng.file-list=/path/to/file/list test-automated

Fixes warning reported by Findbugs.

The data file name stored in the header is not updated if the header and
data files are renamed.  If the stored data file name does not exist,
then a file whose name is similar to the header file's name will be used
(if such a file is present).

Image files can only be detected if a corresponding .hdr file exists, which
won't work if either file was renamed.

Conflicts:
	components/formats-gpl/src/loci/formats/in/CellomicsReader.java

Repository: openmicroscopy/bioformats
Already up-to-date.

Generated by BIOFORMATS-DEV-latest-version-bump#73 (https://ci.openmicroscopy.org/job/BIOFORMATS-DEV-latest-version-bump/73/)

See https://trello.com/c/hTxcpCtl/37-full-repository-skipped-filesets

Update POM version numbers to 5.9.0-SNAPSHOT

…id is not

See https://trello.com/c/3AHJeUPc/236-czi-compressed-pyramidal-error

switch modifier order to "static final" in readers

See https://trac.openmicroscopy.org/ome/ticket/13196

…uested

Since close is called in initFile, this prevents any reader
extending MinimalTiffReader from setting it to a non-default
value.

…lementation

…e calculation

…ze calculation

…Configuration

build: Never run upgrade checker when running unit tests

…late

The file passed to setId is now first on the list, and the directory
listing is sorted (as for other readers).

See openmicroscopy/data_repo_config#299

Memoizer.generateMemo(String)

cachegen can process a single file, a directory, or a text file containing a
list of files.  Directory processing is less sophisticated that the
import candidates style used by the data tests, but this could be
considered for future work.  At the moment, Memoizer is the only reader
in the stack, but FileStitcher could also be added as an option.

This allows the existing skipSave variable to be changed, so that
deleting and saving can be disabled.

This prevents the test from overwriting pre-generated memo files when
they are in a writeable directory.

Trestle: ensure consistent ordering of used files

Support for unconfigured tests

Conflicts:
	components/test-suite/build.xml

In ND files referencing RGB images, sizeC was getting set to 1 rather than 3.
(The individual STK/TIFF files are properly handled). This change multiplies
the sizeC by 3 for RGB images, and replaces a later use of sizeC with
effectiveSizeC.

Fix Metamorph RGB series channel count

Reorder the dimensions for retrieving filenames when splitting

See https://stackoverflow.com/questions/7905501/get-rid-of-pom-not-found-warning-for-org-eclipse-m2elifecycle-mapping/23707050#23707050

This version of the imagej jar includes the needed Java 9 fixes.

Not provided by the Java 9 JVM unless you explicitly opt into it
by enabling the appropriate module.  This works on all Java
versions.

Build updates

Cellomics: fix indexing when the field counts are variable

GelReader: Don't allow overflow when reading unsigned integer values

Backported TiffParser fixes and getIFDs deprecation

http://rsbweb.nih.gov/ij/developer/api/ is now redirecting to
https://imagej.nih.gov/ij/developer/api/.

ND2: prevent integer overflow when reading a tile from a large image

See https://trello.com/c/Zp2rgbqf/246-leica-lif-dimension-problems

All should now be consistent with the existing targets, e.g.
test-automated.

Zeiss CZI: fix big image edge case when tiling is present but a pyramid is not

TiffWriter: Add support for DEFLATE (zlib) compression

…INIMUM

See https://central.sonatype.org/articles/2018/May/04/discontinue-support-for-tlsv11-and-below/ for more details

See https://trello.com/c/zgKbVcun/32-add-documentation-for-getusedfiles-ordering