View source for Pixet 1.8.5 → 1.8.6

== Pixet Python API ==
This release modernises the Python API. Scripts written against 1.8.5 will need changes — the list of what to fix is in §2, with before/after for each item.
----

=== 1. At a glance ===
* Errors are Python exceptions now. Calls raise a <code>PixetError</code> (a subclass by the kind of the error - <code>DeviceError</code>, <code>PixetFileError</code>, <code>PixetValueError</code>, ...) instead of returning <code>-1</code>.
* No more manual memory management. <code>destroy()</code> and the <code>*RefInc()</code> call variants are gone; objects are released for you.
* Common getters became properties — <code>dev.params</code>, <code>dev.pixcfg</code>, <code>dev.dacs</code>, <code>pixet.data_mgr</code>.
* Parameter access is type-free — one <code>getValue()</code> / <code>setValue()</code> pair instead of 39 typed accessors.
* Pixel masks are split into a factory mask, your own user mask, and a read-only combined view.
* <code>pygui</code> is a real widget toolkit — constructors instead of factory methods, plus tables, trees, scroll areas, date/time editors, a reworked plot, and Qt enums.
* Clustering and spectral imaging are proper modules with object APIs and progress callbacks.
* Type stubs ship with every module, so editors give you autocompletion and signature help (with the gaps noted in §3).
* Ten helper modules were removed (§4).

----

=== 2. Breaking changes — what to fix in your scripts ===

==== 2.1 Check for exceptions, not return codes ====
Failures now raise a <code>pypixet.PixetError</code>; bad arguments raise the usual <code>TypeError</code> / <code>ValueError</code>. The subclass tells the kind of the error - <code>DeviceError</code> (a failed device method, acquisition or communication), <code>PixetFileError</code> (also an <code>OSError</code>), <code>PixetTimeoutError</code> (also a <code>TimeoutError</code>), <code>PixetValueError</code> (also a <code>ValueError</code>) and <code>PixetNotFoundError</code> (also a <code>LookupError</code>). Every exception has <code>code</code> (an <code>ErrorCode</code> member) and <code>message</code>. The classes exist as soon as <code>pypixet</code> is imported. <code>pypixet.start()</code> and <code>pypixet.exit()</code> themselves raise plain <code>RuntimeError</code> — they run before the Pixet API is up.
 # 1.8.5
 rc = dev.doSimpleAcquisition(1, 0.1, pixet.PX_FTYPE_AUTODETECT, "out.pmf")
 if rc < 0:
     print("acquisition failed:", pypixet.getLasError())
 
 # 1.8.6
 try:
     dev.doSimpleAcquisition(1, 0.1, "out.pmf")
 except pypixet.PixetError as exc:
     print("acquisition failed:", exc)
<code>pypixet.getLasError()</code> has been removed — the message is on the exception.

==== 2.2 Delete every <code>destroy()</code> and <code>*RefInc()</code> call ====
 # 1.8.5
 frame = dev.lastAcqFrameRefInc()
 data = frame.data()
 frame.destroy()
 
 # 1.8.6
 frame = dev.lastAcqFrame()
 data = frame.data()
Removed everywhere: <code>destroy()</code>, <code>destroyCE()</code>, <code>incRefCount()</code>, <code>decRefCount()</code>, <code>privateInterface()</code>, and the <code>RefInc</code> variants <code>acqDataRefInc</code>, <code>acqFrameRefInc</code>, <code>lastAcqDataRefInc</code>, <code>lastAcqFrameRefInc</code>, <code>lastAcqPixelsRefInc</code>, <code>lastAcqClustersRefInc</code>.

==== 2.3 Device parameters: one generic accessor ====
<code>dev.parameters()</code> returned a parameter manager with 13 typed create/get/set families. It is replaced by the <code>dev.params</code> property and a single value API.
 # 1.8.5
 bias = dev.parameters().get("BiasVoltage").getDouble()
 dev.parameters().get("BiasVoltage").setDouble(200.0)
 
 # 1.8.6
 bias = dev.params.getValue("BiasVoltage")
 dev.params.setValue("BiasVoltage", 200.0)
<code>getValue()</code> returns the natural Python type. If you need to know it, <code>dev.params.type("BiasVoltage")</code> returns an <code>int</code> you can compare against the new <code>pixet.DataType</code> enum (<code>Double</code>, <code>String</code>, …), which is an <code>IntEnum</code> — <code>dev.params.type(name) == pixet.DataType.DOUBLE</code>. Blobs come back through <code>getBlob()</code>, and <code>isBlob()</code> tells you when to use it. <code>addValue(name, value, description=<nowiki>''</nowiki>, dataType=-1)</code> replaces <code>addBOOL</code> / <code>addI32</code> / <code>addString</code> and the rest.

==== 2.4 Getters that became properties ====
 dev.parameters()      →  dev.params
 dev.pixCfg()          →  dev.pixcfg
 dev.dacs()            →  dev.dacs
 pixet.dataMgr()       →  pixet.data_mgr
Drop the parentheses — these are attributes, not calls. <code>dacs</code> keeps its name and only loses the parentheses; the rest of the DAC api changed as well, see §2.8.

==== 2.5 Acquisition results ====
 dev.lastAcqFrameRefInc()    →  dev.lastAcqFrame()
 dev.lastAcqDataRefInc()     →  dev.lastAcqData()
 dev.acqFrameRefInc(i)       →  dev.acqFrame(i)
 dev.acqDataRefInc(i)        →  dev.acqData(i)
<code>doSimpleAcquisition()</code> has no <code>fileType</code> argument any more - the file type is always detected from the file name (<code>PX_FTYPE_AUTODETECT</code>) - and <code>fileName</code> defaults to <code><nowiki>''</nowiki></code>, so <code>dev.doSimpleAcquisition(1, 0.1)</code> is enough for a measurement you only want in memory.

New on Timepix3 devices:
 dev.lastAcqClustersRefInc()   →  dev.<code>lastAcqClusters</code>()
 dev.lastAcqPixelsRefInc()     →  dev.lastAcqPixels()

==== 2.6 <code>save()</code> takes the data manager ====
 # 1.8.5
 frame.save("out.pmf", pixet.PX_FTYPE_AUTODETECT)
 
 # 1.8.6
 frame.save(pixet.data_mgr, "out.pmf", pixet.FileType.AUTODETECT)
<code>load()</code> is unchanged.

==== 2.7 Pixel masking: factory mask vs user mask ====
The single mask was split into a factory mask (from the detector's calibration) and a user mask (yours), plus a read-only combined view. Every <code>mask*</code> name gained a <code>Factory</code> or <code>User</code> qualifier.
 # 1.8.5
 dev.pixCfg().mask(pixel, True)
 n = dev.pixCfg().maskedCount()
 m = dev.pixCfg().maskMatrix()
 
 # 1.8.6
 dev.pixcfg.maskUser(pixel, True)
 n = dev.pixcfg.userMaskedCount()          # or combinedMaskedCount()
 m = dev.pixcfg.userMaskMatrix()           # or combinedMaskMatrix()
<code>maskUser()</code> takes a linear pixel index, as <code>mask()</code> did; <code>maskUserRect(x, y, width, height, masked)</code> is the one that works in coordinates.

The same rule applies to the chip, row, column and rect variants (<code>maskUserChip</code>, <code>maskUserRow</code>, <code>maskUserColumn</code>, <code>maskUserRect</code>, <code>isUserMasked*</code>, <code>setUserMaskMatrix*</code>, …) and to their <code>Factory</code> counterparts. Also new: <code>clearUserMask()</code>, <code>clearFactoryMask()</code>, <code>copyUserMask()</code>, <code>copyMasksFromPixCfg()</code>, <code>isCombinedMasked()</code>.

On Medipix2/Timepix pixel configs, the whole-matrix <code>modeMatrix()</code> / <code>setModeMatrix()</code> were removed; the per-chip <code>modeMatrixChip()</code> / <code>setModeMatrixChip()</code> and the <code>setMode*</code> calls are unchanged.

==== 2.8 DACs: a property, and <code>copy()</code> instead of <code>createCopyOfDacs()</code> ====
 # 1.8.5
 dacs = dev.dacs()                     # the DACs of the device
 copy = dev.createCopyOfDacs()         # an independent set with the values of the device
 ...
 copy.destroy()
 
 # 1.8.6
 dacs = dev.dacs                       # a property, the same object on every access
 copy = dacs.copy()                    # an independent set with the values of 'dacs'
<code>dev.createCopyOfDacs()</code> no longer exists — <code>dev.dacs.copy()</code> replaces it, and <code>copy.copy(dev.dacs)</code> / <code>copy.deepcopy(dev.dacs)</code> do the same. Both hand back a set which already carries the values it was copied from and which nothing has to destroy. The difference is what is copied: <code>createCopyOfDacs()</code> always took the DACs of the device, while <code>copy()</code> takes the set you call it on — <code>dev.dacs_opm["CSM"].copy()</code> copies that operation mode. <code>copyFromDacs()</code> is still there for copying between two sets you already have.

<code>dacs.is_main</code> tells the DACs of the device (<code>1</code>) from a copy (<code>0</code>), and writing a copy leaves the device untouched until you send the set to it.

Setting values is unchanged: <code>dacs.setDac(index, chip, value)</code>, <code>dacs.setToDevice()</code>, <code>dacs.setDefault()</code>, <code>dev.setDacs(dacs)</code>, <code>dev.refreshDacs()</code>. <code>dacs.names()</code> returns the DAC names as a list of <code>str</code>, and <code>thresholdToDac()</code> / <code>dacToThreshold()</code> convert against the chip type of the device.

<code>dev.createCopyOfPixCfg()</code> went the same way — <code>dev.pixcfg.copy()</code>.

Medipix3 operation modes
 # 1.8.5
 dacs = dev.dacsMode(pixet.PX_MPX3_OPM_CSM)
 copy = dev.createCopyOfDacsMode(pixet.PX_MPX3_OPM_CSM)
 dev.setDacsMode(pixet.PX_MPX3_OPM_CSM, dacs)
 
 # 1.8.6
 dacs = dev.dacs_opm[pixet.Mpx3OpMode.CSM]     # or dev.dacs_opm["CSM"]
 copyDacs = dacs.copy()
 copyDacs.setToDevice()                        # dev.setDacsMode(pixet.Mpx3OpMode.CSM, copyDacs)
<code>dacsMode()</code> and <code>createCopyOfDacsMode()</code> are removed. <code>dev.dacs_opm</code> is a sequence of the <code>IMpxDacs</code> of every operation mode - <code>len()</code>, iteration, <code>names()</code>, and an index which is the mode number or its name (case insensitive, the <code>PX_MPX3_OPM_</code> prefix optional). The object of a mode is cached, so <code>dev.dacs_opm[0] is dev.dacs_opm[0]</code>, and modes which share one set of DACs in the device share the object. <code>dev.pixcfg_opm</code> is the same for the pixel configurations, in place of <code>pixCfgMode()</code> / <code>createCopyOfPixCfgMode()</code>.

==== 2.9 Type casting: check the data format ====
The wrapper classes now form a real hierarchy — <code>IDevMpx3</code> derives from <code>IDevMpx</code> derives from <code>IDev</code>, <code>IMpxFrame</code> derives from <code>IData</code>. Most of the time no cast is needed at all: <code>dev.lastAcqFrame()</code> and <code>dev.acqFrame(i)</code> already hand you an <code>IMpxFrame</code>.
 # 1.8.5
 frame = data.asIMpxFrame()
 if data.canCastToDataFormat(pixet.PX_DATAFORMAT_FRAME):
     ...
 
 # 1.8.6
 if data.dataFormat() == pixet.DataFormat.FRAME:
     ...
Removed: <code>asIData()</code>, <code>asIMpxFrame()</code>, <code>asIDev()</code>, <code>canCastToDataFormat()</code>.

The wrapper types themselves are not module attributes — there is no <code>pypixet.IMpxFrame</code> to pass to <code>isinstance()</code>. Test the data format as above, or compare <code>type(data).__name__</code> with <code>"IMpxFrame"</code>.

==== 2.10 Other renamed calls ====
{| class="wikitable"
!1.8.5
!1.8.6
|-
|<code>frame.convertToChipByChip()</code>
|<code>frame.chipByChipImage()</code>
|-
|<code>frame.ignoreMasked()</code>
|<code>frame.setIgnoreMasked()</code>
|}
The plugin API is reduced, not removed: <code>pixet.plugins()</code> still lists the loaded plugins and <code>IPlugin</code> keeps <code>id()</code>, <code>name()</code>, <code>execute()</code>, <code>params</code>, <code>registerEvent()</code> and <code>unregisterEvent()</code>. Gone from it are <code>pixet.addPlugin()</code>, <code>pixet.plugin(name)</code> and the <code>IMenuItem</code> type.

Also removed with no replacement in this release: the filter config-dialog hooks (<code>hasConfigGUI()</code>, <code>showConfigGUI()</code>), custom file-type registration (<code>dataMgr.registerFileType()</code> / <code>unregisterFileType()</code>), and the <code>IByteBuff</code> / <code>IRefByteBuff</code> buffer objects — buffer data is passed as plain Python lists.

==== 2.11 <code>pygui</code>: widgets are classes ====
If you build panels with <code>pygui</code>, the factory methods are gone. Construct widgets directly.
 # 1.8.5
 app = pygui.App(sys.argv)
 win = pygui.newMainWindow()
 btn = win.newPushButton("Start")
 plot = win.newPlot()
 btn.setEventHandler(on_click)
 
 # 1.8.6
 app = pygui.Application(sys.argv)
 win = pygui.MainWindow()
 btn = pygui.PushButton("Start", win)
 plot = pygui.Plot(win)
 btn.eventFilter().setHandler(on_click)

* All 24 <code>newXxx()</code> factory methods are replaced by constructors of the same name minus <code>new</code> — except <code>newButton()</code>, which was an alias of <code>newPushButton()</code>; use <code>PushButton</code>.
* <code>setEventHandler()</code> / <code>setFilterEvents()</code> / <code>updatefilterEvents()</code> / <code>getfilterEvents()</code> are replaced by <code>eventFilter()</code>, which gives the one <code>EventFilter</code> object of the widget - <code>handler()</code> / <code>setHandler()</code>, <code>events()</code> / <code>setEvents()</code> and <code>isEnabled()</code> / <code>setEnabled()</code> are its methods.
* <code>App()</code> became the <code>Application(argv)</code> constructor.
* The remaining application module functions moved onto <code>Application</code> as static methods — <code>appExists()</code> is <code>Application.exists()</code>, and <code>processEvents()</code>, <code>getMainWindow()</code>, <code>screens()</code>, <code>screenGeometry()</code>, <code>setCursor()</code>, <code>cursorShape()</code>, <code>beep()</code> keep their names (<code>pygui.Application.processEvents()</code>, …). <code>loadFromUiFile()</code> / <code>loadFromUiString()</code> stay module functions.
* The <code>pygui.app</code> module attribute is gone — keep the <code>Application</code> object you construct.
* <code>WPlot</code> is now <code>Plot</code>. <code>setRangeZ()</code> was removed; layers, lines, scatters, highlights and the axes are returned as objects instead of integer indices. A line, a bar line and a highlight are added on the layer (<code>plot.addLayer().addLine()</code>, <code>plot.layer(0).addHighlight()</code>), and the index based duplicates of the object methods are gone (<code>setLineColor()</code>, <code>setLineWidth()</code>, <code>setScatter*()</code>, <code>setHighlight*()</code>, <code>rangeX/Y()</code>, <code>setRangeX/Y()</code>, <code>setAutoRangeX/Y()</code>, <code>setScaleTypeX/Y()</code>, the counts, <code>clearLayer/Line/Scatter()</code>, <code>showLine/Scatter()</code>). <code>plot.setLineColor(0, 0, 'red')</code> is <code>plot.layer(0).addLine('red')</code>, <code>plot.setRangeX(0, 10)</code> is <code>plot.axisX.setRange(0, 10)</code>. On the layer <code>setLineData()</code>, <code>addLinePoint()</code>, <code>replaceLinePoint()</code> and <code>addScatterPoint()</code> keep taking indices.
* <code>MpxFrame.setAutoResize()</code> was removed.

==== 2.12 Constants became enum classes ====
The <code>PX_*</code>, <code>PLUGIN_*</code> and <code>PLGERR_*</code> constants are no longer attributes of <code>pixet</code>. Each group is an enum class of the <code>pypixet</code> module, <code>pixet</code> holds the same classes (<code>pypixet.FileType is pixet.FileType</code>). The member name is the constant without its group prefix, in upper case (<code>PX_FTYPE_PNG</code> → <code>FileType.PNG</code>, <code>PX_CE_INIT_DEVICE</code> → <code>CeCmd.INIT_DEVICE</code>).
 # 1.8.5
 frame.save("out.png", pixet.PX_FTYPE_PNG)
 
 # 1.8.6
 frame.save("out.png", pixet.FileType.PNG)
<code>IntEnum</code> / <code>IntFlag</code> members still compare equal to the old numbers and <code>StrEnum</code> members to the old strings, so a plain <code>int</code> or <code>str</code> is accepted wherever a member is.
{| class="wikitable"
!Constants
!Enum class
!Example
|-
|<code>PX_PAR_*</code>
|<code>ParamFlag</code>
|<code>PX_PAR_READONLY</code> → <code>ParamFlag.READONLY</code>
|-
|<code>PX_ATTR_*</code>
|<code>ParamAttr</code>
|<code>PX_ATTR_READONLY</code> → <code>ParamAttr.READONLY</code>
|-
|<code>PX_PAR_VALUE_*</code>
|<code>ParamChange</code>
|<code>PX_PAR_VALUE_CHANGED</code> → <code>ParamChange.CHANGED</code>
|-
|<code>PX_PARAM_*</code>
|<code>PixetParam</code>
|<code>PX_PARAM_APIVERSION</code> → <code>PixetParam.APIVERSION</code>
|-
|<code>PX_DATAFORMAT_*</code>
|<code>DataFormat</code>
|<code>PX_DATAFORMAT_FRAME</code> → <code>DataFormat.FRAME</code>
|-
|<code>PX_FTYPE_*</code>
|<code>FileType</code>
|<code>PX_FTYPE_AUTODETECT</code> → <code>FileType.AUTODETECT</code>
|-
|<code>PX_DATAEVENT_*</code>
|<code>DataEventType</code>
|<code>PX_DATAEVENT_BEFORE_SAVE</code> → <code>DataEventType.BEFORE_SAVE</code>
|-
|<code>PX_FRAMETYPE_*</code>
|<code>FramePixelType</code>
|<code>PX_FRAMETYPE_I16</code> → <code>FramePixelType.I16</code>
|-
|<code>PX_FRAMESAVE_*</code>
|<code>SaveFlag</code>
|<code>PX_FRAMESAVE_BINARY</code> → <code>SaveFlag.BINARY</code>
|-
|<code>PX_MFFORMAT_*</code>
|<code>MultiFrameFormat</code>
|<code>PX_MFFORMAT_BINARY_MULTIFRAME</code> → <code>MultiFrameFormat.BINARY_MULTIFRAME</code>
|-
|<code>PX_DATAFILTER_*</code>
|<code>DataFilterType</code>
|<code>PX_DATAFILTER_NORMAL</code> → <code>DataFilterType.NORMAL</code>
|-
|<code>PX_DEVINFO_VENDOR_*</code>
|<code>Vendor</code>
|<code>PX_DEVINFO_VENDOR_IEAP</code> → <code>Vendor.IEAP</code>
|-
|<code>PX_DEVINFO_TYPE_*</code>
|<code>DevModel</code>
|<code>PX_DEVINFO_TYPE_DUMMY</code> → <code>DevModel.DUMMY</code>
|-
|<code>PX_DEVTYPE_*</code>
|<code>DevType</code>
|<code>PX_DEVTYPE_MPX2</code> → <code>DevType.MPX2</code>
|-
|<code>PX_DEVSTATUS_*</code>
|<code>DevStatus</code>
|<code>PX_DEVSTATUS_OK</code> → <code>DevStatus.OK</code>
|-
|<code>PX_EVENT_*</code>
|<code>EventName</code>
|<code>PX_EVENT_BUSY_CHANGED</code> → <code>EventName.BUSY_CHANGED</code>
|-
|<code>PX_PAR_EVENT_VALUE_CHANGED</code>
|<code>EventName</code>
|<code>PX_PAR_EVENT_VALUE_CHANGED</code> → <code>EventName.PARAMETER_VALUE_CHANGED</code>
|-
|<code>PX_EVENT_XRAY_STATUS_*</code>
|<code>XrayEventName</code>
|<code>PX_EVENT_XRAY_STATUS_XRAYON</code> → <code>XrayEventName.XRAYON</code>
|-
|<code>PX_MSGTYPE_*</code>
|<code>LogMsgType</code>
|<code>PX_MSGTYPE_INFO</code> → <code>LogMsgType.INFO</code>
|-
|<code>PX_ACQTYPE_*</code>
|<code>AcqType</code>
|<code>PX_ACQTYPE_FRAMES</code> → <code>AcqType.FRAMES</code>
|-
|<code>PX_ACQMODE_*</code>
|<code>AcqMode</code>
|<code>PX_ACQMODE_NORMAL</code> → <code>AcqMode.NORMAL</code>
|-
|<code>PX_REPEAT_*</code>
|<code>RepeatFlags</code>
|<code>PX_REPEAT_OVERWRITE_SERIES</code> → <code>RepeatFlags.OVERWRITE_SERIES</code>
|-
|<code>PX_CHIPTYPE_*</code>
|<code>ChipType</code>
|<code>PX_CHIPTYPE_MXR</code> → <code>ChipType.MXR</code>
|-
|<code>PX_PIXCFG_MXR</code>, … (chip types)
|<code>PixCfgType</code>
|<code>PX_PIXCFG_MXR</code> → <code>PixCfgType.MXR</code>
|-
|<code>PX_PIXCFG_ALL</code>, … (config items)
|<code>PixCfgPart</code>
|<code>PX_PIXCFG_ALL</code> → <code>PixCfgPart.ALL</code>
|-
|<code>PX_THLFLG_*</code>
|<code>ThlFlag</code>
|<code>PX_THLFLG_NONE</code> → <code>ThlFlag.NONE</code>
|-
|<code>PX_MPX3_OPM_*</code>
|<code>Mpx3OpMode</code>
|<code>PX_MPX3_OPM_SPM_1CH</code> → <code>Mpx3OpMode.SPM_1CH</code>
|-
|<code>PX_MPX3_CNT_*</code>
|<code>Mpx3Counter</code>
|<code>PX_MPX3_CNT_LOW</code> → <code>Mpx3Counter.LOW</code>
|-
|<code>PX_MPX3_CNTD_*</code>
|<code>Mpx3Depth</code>
|<code>PX_MPX3_CNTD_12B</code> → <code>Mpx3Depth.BITS_12</code>
|-
|<code>PX_MPX3_COLB_*</code>
|<code>Mpx3ColBlock</code>
|<code>PX_MPX3_COLB_ALL</code> → <code>Mpx3ColBlock.ALL</code>
|-
|<code>PX_MPX3_ROWB_*</code>
|<code>Mpx3RowBlock</code>
|<code>PX_MPX3_ROWB_ALL</code> → <code>Mpx3RowBlock.ALL</code>
|-
|<code>PX_MPX3_GAIN_*</code>
|<code>Mpx3Gain</code>
|<code>PX_MPX3_GAIN_SUPERHIGH</code> → <code>Mpx3Gain.SUPERHIGH</code>
|-
|<code>PX_TPX3_OPM_*</code>
|<code>Tpx3OpMode</code>
|<code>PX_TPX3_OPM_TOATOT</code> → <code>Tpx3OpMode.TOATOT</code>
|-
|<code>PX_TPX3_GRAYCNT_*</code>
|<code>Tpx3GrayCounter</code>
|<code>PX_TPX3_GRAYCNT_AUTO</code> → <code>Tpx3GrayCounter.AUTO</code>
|-
|<code>PX_TPX3_FRAMETYPE_*</code>
|<code>Tpx3FrameType</code>
|<code>PX_TPX3_FRAMETYPE_SUM</code> → <code>Tpx3FrameType.SUM</code>
|-
|<code>PX_TPX2_OPM_*</code>
|<code>Tpx2OpMode</code>
|<code>PX_TPX2_OPM_TOT10_TOA18</code> → <code>Tpx2OpMode.TOT10_TOA18</code>
|-
|<code>PX_STGTYPE_*</code>
|<code>SettingsFormat</code>
|<code>PX_STGTYPE_INI</code> → <code>SettingsFormat.INI</code>
|-
|<code>PLUGIN_ID_*</code>
|<code>PluginId</code>
|<code>PLUGIN_ID_DEV_FEEDER</code> → <code>PluginId.DEV_FEEDER</code>
|-
|<code>PLUGIN_EVENT_*</code>
|<code>PluginEvent</code>
|<code>PLUGIN_EVENT_INITIALIZED</code> → <code>PluginEvent.INITIALIZED</code>
|-
|<code>PLGERR_*</code>
|<code>PluginErrorId</code>
|<code>PLGERR_NO_ERROR</code> → <code>PluginErrorId.NO_ERROR</code>
|-
|<code>PXERR_*</code> (<code>pxerrors.h</code>)
|<code>ErrorCode</code>
|<code>PXERR_FILE_OPEN</code> → <code>ErrorCode.FILE_OPEN</code>
|-
|<code>PX_MXR_*</code> (Medipix2 MXR DACs)
|<code>MxrDacs</code>
|<code>PX_MXR_IKRUM</code> → <code>MxrDacs.IKRUM</code>
|-
|<code>PX_TPX_*</code> (Timepix DACs)
|<code>TpxDacs</code>
|<code>PX_TPX_IKRUM</code> → <code>TpxDacs.IKRUM</code>
|-
|<code>PX_MPX3_*</code> (Medipix3 DACs)
|<code>Mpx3Dacs</code>
|<code>PX_MPX3_TH0</code> → <code>Mpx3Dacs.TH0</code>
|-
|<code>PX_TPX3_*</code> (Timepix3 DACs)
|<code>Tpx3Dacs</code>
|<code>PX_TPX3_IBIAS_PREAMP_ON</code> → <code>Tpx3Dacs.IBIAS_PREAMP_ON</code>
|-
|<code>PX_TPX2_*</code> (Timepix2 DACs)
|<code>Tpx2Dacs</code>
|<code>PX_TPX2_PREAMP_ON</code> → <code>Tpx2Dacs.PREAMP_ON</code>
|-
|<code>PX_TPX2_DACOUT_*</code>
|<code>Tpx2DacOuts</code>
|<code>PX_TPX2_DACOUT_DOUT_PREAMP_ON</code> → <code>Tpx2DacOuts.DOUT_PREAMP_ON</code>
|}
A class is created only when a method, an argument or a struct field uses it. No method uses <code>PX_DFUID_*</code>, <code>PX_EXT_*</code>, <code>PX_METADATA_*</code>, <code>PX_SENSOR_TYPE_*</code>, <code>PX_TPXMODE_*</code>, <code>PX_MPX2_DOUTPUT_*</code>, <code>PX_DEVMPX2_CALIB_*</code>, <code>PX_TPX3_VCO_*</code>, <code>PX_PIXETTYPE_*</code> and <code>PX_STATUS_*</code>, so they are no longer available in Python - use the plain value.

Only four values stay plain attributes of <code>pixet</code>: <code>PX_PIXET_VERSION</code> → <code>pixet.VERSION</code>, <code>PX_CHIP_ALL</code> → <code>pixet.CHIP_ALL</code>, <code>PX_TPX3_PIXSAVE_FILTERDUMMY</code> → <code>pixet.TPX3_PIXSAVE_FILTERDUMMY</code> and <code>PX_TPX3_PIXADD_UPDATE_PREVIEW</code> → <code>pixet.TPX3_PIXADD_UPDATE_PREVIEW</code>.
----

=== 3. What's new ===

==== Detector control ====
* <code>pixet.devicesTpx2()</code> joins the existing per-type device lists.
* Device log events: <code>dev.registerLogEvent()</code> / <code>unregisterLogEvent()</code>.
* <code>registeredEvents()</code> lists what <code>registerEvent()</code> has registered on an object - one <code>EventItem</code> per registration, with the read only members <code>name</code>, <code>callbackId</code> and <code>func</code> (the registered callable). On <code>IDev</code>, <code>IDev2</code>, <code>IPixet</code>, <code>IParamSet</code> and <code>IPlugin</code> - everything that registers events. Log events have their own handler and are not in the list.
* Constants are enum classes. <code>pixet.PX_FTYPE_AUTODETECT</code> and friends were invisible to <code>dir()</code> and to editors. They are now members of enum classes of the module - <code>pypixet.FileType.AUTODETECT</code>, <code>pypixet.DataType.DOUBLE</code> - listed in the stub, so autocompletion finds them, and the methods return members (<code>dev.deviceType()</code> is <code><DevType.TPX3: 3></code>).

==== <code>pygui</code> ====
New widgets and types: <code>Table</code>, <code>TreeWidget</code>, <code>ScrollArea</code>, <code>Frame</code>, <code>DateEdit</code>, <code>TimeEdit</code>, <code>DateTimeEdit</code>, <code>BoxLayout</code>, <code>FormLayout</code>, and the value types <code>Point</code>, <code>Size</code>, <code>Rect</code>, <code>Color</code>. Qt constants are exposed as proper enums — <code>CursorShape</code>, <code>WindowType</code>, <code>EventType</code>, <code>AligmentFlag</code>, <code>ScrollBarPolicy</code>, <code>LayoutSizeConstraint</code>, <code>FrameShape</code>, <code>FrameShadow</code> and more.

<code>Plot</code> gains per-object layers, lines, scatters, highlights and axes (<code>PlotAxis</code> on <code>axisX</code>, <code>axisY</code>, <code>secAxisX</code>, <code>secAxisY</code>), log/linear axis scales, colour maps, the nested <code>ScaleType</code> / <code>AxisType</code> / <code>AxisOrientation</code> / <code>LabelType</code> / <code>ColorMap</code> enums, and <code>setLineData()</code> for updating a curve in one call. <code>MpxFrame</code> / <code>MpxFramePanel</code> gain <code>data()</code>, <code>setGamma()</code>, zoom control (<code>setZoom()</code>, <code>setZoomRect()</code>, <code>zoomFactor()</code>, <code>zoomOut()</code>), selection rectangles and a <code>toolTipCallback</code> member, plus the nested <code>SizeType</code> / <code>MarkType</code> enums. <code>ComboBox</code> and <code>PropertyTreeView</code> are much more complete.

The dialog module functions are unchanged — <code>showMessage()</code>, <code>showWarning()</code>, <code>showError()</code>, <code>showQuestion()</code>, <code>inputText()</code>, <code>showItemSelection()</code>, <code>getOpenFileName()</code>, <code>getSaveFileName()</code>, <code>getSaveFileNameAndFilter()</code>, <code>getExistingDirectory()</code> all keep their 1.8.5 names and signatures.

==== <code>pyclustering</code> ====
A real module rather than a plugin shim, with <code>Pixel</code> and <code>Cluster</code> value types:
 import pypixet
 import pyclustering
 
 pixet = pypixet.start()
 device = pixet.devices()[0]
 
 c = pyclustering.Clustering(device)
 c.newClustersCb = lambda clusters, acqIndex: print(len(clusters), "clusters")
 c.loadCalibrationFromDevice()
 c.startMeasurement(1.0, 60.0, "run.clog")
<code>Clustering(device)</code> takes the device the measurement runs on; without it only <code>replayData()</code> works — <code>startMeasurement()</code> and <code>loadCalibrationFromDevice()</code> need one. The methods take positional arguments only, keyword arguments raise <code>TypeError</code>.

Also: <code>replayData()</code>, <code>abort()</code>, <code>isRunning()</code>, <code>loadCalibrationFromFiles()</code>, <code>isCalibrationLoaded()</code>, and the callbacks <code>messageCb</code>, <code>progressCb</code>, <code>newClustersWithPixelsCb</code>, <code>acqStartedCb</code>, <code>acqFinishedCb</code>.

==== <code>pyspectraimg</code> ====
Replaces the flat <code>si*</code> functions with a <code>SpectraImg</code> object and a <code>ProcParameters</code> settings object:
 import pyspectraimg
 
 si = pyspectraimg.SpectraImg(dev)
 si.progressCb = lambda finished, pct: print(f"{pct:.0f}%")
 si.loadCalibrationFromDevice()
 si.startMeasurement(1.0, 60.0, "run.h5", True)
 img = si.getFrameForEnergyRange(10, 40, True)
The device argument is mandatory — <code>SpectraImg</code> reads the matrix size and the chip layout from it. As in <code>pyclustering</code>, the methods take positional arguments only.

Plus <code>getGlobalSpectrum()</code>, <code>getGlobalSpectrumInRect()</code>, <code>saveSumFrame()</code>, <code>saveDataAsFramesToFile()</code>, <code>setXrfParams()</code>, <code>processedPixelsPerSecond()</code>, and XRF correction options on <code>ProcParameters</code>.

==== <code>pyutils</code> ====
* New <code>linearRegression(y, x)</code>.
* Results are named tuples now, so <code>result.mean</code> works alongside <code>result[0]</code>: <code>SpectrumPeak</code>, <code>GerfFitResult</code>, <code>FitPeakResult</code>, <code>PerformToTfitResult</code>, <code>LinearRegressionResult</code>, <code>ClustersWithParams</code>.
* <code>calibGetSpectrumPeak()</code> was renamed to <code>calibSpectrumPeak()</code>.
* <code>performToTfit()</code> no longer takes a <code>stddev_out</code> argument — the value is in the result.

==== <code>pyftdi</code> ====
The module-level functions now raise <code>RuntimeError</code> with the FTDI error text instead of returning a status code, and <code>open()</code> closes a device left open by an earlier <code>open()</code>. The <code>Ftdi</code> class — which already existed in 1.8.5 and lets you drive several devices at once — keeps its status-code return values, so its methods are unchanged.
 # 1.8.5
 rc = pyftdi.open("FT4232H", 115200)
 if rc != 0:
     ...
 
 # 1.8.6
 try:
     pyftdi.open("FT4232H", 115200)
 except RuntimeError as exc:
     ...

==== Editor support ====
Every module ships a <code>.pyi</code> type stub — <code>pypixet</code>, <code>pygui</code>, <code>pyutils</code>, <code>pyclustering</code>, <code>pyspectraimg</code>, <code>pyftdi</code>, <code>pyhid</code>. Editors and type checkers pick these up automatically and give you completion, parameter hints and type checking against the API.

The stubs are generated, and two of them are behind the sources in this release: <code>pypixet.pyi</code> lists only a few of <code>IPixet</code>'s methods and still declares the removed <code>getLasError()</code>, and <code>pygui.pyi</code> omits the static methods of <code>Application</code>. The modules themselves are correct — use <code>help()</code> / <code>dir()</code> when a stub looks incomplete.
----

=== 4. Removed modules ===
{| class="wikitable"
!Module
!Replacement
|-
|<code>pymeasutils</code>
|<code>pyspectraimg</code> — same functionality, object API
|-
|<code>pybhfilter</code>, <code>pyfffilter</code>
|The filters themselves are unchanged; reach them through <code>pixet.filter_mgr</code> (<code>filters()</code>, <code>filterByName()</code>, <code>addFilter()</code>)
|-
|<code>pypxproc</code>
|No direct replacement
|-
|<code>pydevcontrol</code>
|No direct replacement
|-
|<code>pyequalization</code>
|No direct replacement
|-
|<code>pyfitpix</code>, <code>pyminipix</code>, <code>pymodupix</code>
|Device-specific helpers; use the standard <code>IDev</code> / <code>IDevMpx</code> API
|-
|<code>pypyscripting</code>
|No direct replacement
|}
The underlying Pixet features are unaffected — it is the thin Python wrapper modules that were retired. If you depend on one of these, contact support before upgrading so we can cover your use case.
----

=== 5. Upgrade checklist ===
# Wrap acquisition and configuration calls in <code>try</code> / <code>except pypixet.PixetError</code>; remove <code>-1</code> checks and <code>pypixet.getLasError()</code>.
# Delete every <code>destroy()</code> call and rename <code>*RefInc()</code> to the plain name.
# Replace <code>dev.parameters().get(name).getX()</code> with <code>dev.params.getValue(name)</code>.
# Drop the parentheses from <code>dev.dacs()</code>, <code>dev.pixCfg()</code>, <code>pixet.dataMgr()</code>, <code>pixet.filterMgr()</code>.
# Replace <code>dev.createCopyOfDacs()</code> with <code>dev.dacs.copy()</code> (and <code>dev.dacsMode(m)</code> with <code>dev.dacs_opm[m]</code>).
# Add <code>pixet.data_mgr</code> as the first argument to <code>save()</code>.
# Qualify mask calls with <code>User</code> or <code>Factory</code>.
# Drop <code>asIMpxFrame()</code> — the acquisition getters already return <code>IMpxFrame</code> — and replace <code>canCastToDataFormat()</code> with a <code>data.dataFormat()</code> comparison.
# In <code>pygui</code> code, swap <code>newXxx()</code> factories for constructors and <code>setEventHandler()</code> for <code>eventFilter().setHandler()</code>.
# Replace the <code>pixet.PX_*</code> constants with the enum class members (§2.12).
# Check §4 for any removed module you import.