Fix #1045. Ensure line plot displays integer-valued data items in log display.

[cmeyer]

No description provided.

[ndellby]

Why do we care about integer data? (email by Outlook, if I can find it)

…

-Bruker Confidential-

________________________________ From: Chris Meyer ***@***.***> Sent: Tuesday, May 14, 2024 4:54:01 PM To: nion-software/nionswift ***@***.***> Cc: Subscribed ***@***.***> Subject: [nion-software/nionswift] Fix #1045. Ensure line plot displays integer-valued data items in log display. (PR #1046) **EXTERNAL EMAIL**

________________________________ You can view, comment on, or merge this pull request online at: #1046 Commit Summary * 0358123<0358123> Fix #1045. Ensure line plot displays integer-valued data items in log display. File Changes (2 files<https://github.com/nion-software/nionswift/pull/1046/files>) * M nion/swift/LineGraphCanvasItem.py<https://github.com/nion-software/nionswift/pull/1046/files#diff-fc43e746fd3782b16ec910753de2c1975a18d9ada87e062f480b68fc50f6680d> (4) * M nion/swift/test/LineGraphCanvasItem_test.py<https://github.com/nion-software/nionswift/pull/1046/files#diff-8b5d45db2d3169b494e49234f06bc7d349b445ab42f9b51732c81b39b7e75522> (8) Patch Links: * https://github.com/nion-software/nionswift/pull/1046.patch * https://github.com/nion-software/nionswift/pull/1046.diff — Reply to this email directly, view it on GitHub<#1046>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ALBPXD3VLOLNIPF2RVBU3NLZCKPZTAVCNFSM6AAAAABHXDGOFOVHI2DSMVQWIX3LMV43ASLTON2WKOZSGI4TMNJZHAYDOMA>. You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[cmeyer]

Why do we care about integer data? (email by Outlook, if I can find it)

Data items can contain numpy arrays of any data type, including integer. The line plot was not displaying and showing a stack trace.

[Tiomat85]

Fix makes sense, forcing it to float if it isn't.
Question on your unit test though, you are asserting the min is <0.0, yet your lowest data point is 1.

[cmeyer]

Fix makes sense, forcing it to float if it isn't. Question on your unit test though, you are asserting the min is <0.0, yet your lowest data point is 1.

I added some explanation to the code (force pushed). Rough answer: the returned values are log10 axis values, so will be slightly broader range than the original data and log10. i.e. 1,2,3 data values --> 0.98 -> 4 axis limits --> log10(0.98) -> log10(4). log10(0.98) < 0.0; log10(4) > 0.0. These are only very rough sanity checks; the test is really just to call the calculate function in the first place. The actual values for various scenarios are checked in different unit tests.