New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
implement depends_on #273
Comments
Original reporter: prjemian We need to decide whether this is an attribute or a field and use just one of them. On the wiki page (above), NXdetector_element, is used both ways. Looking at how it applies in all cases shown on the wiki, it should be an attribute of both groups and fields rather than defined as a field. There are obvious advantages to defining "depends_on" as an attribute:
|
Original reporter: prjemian Please describe how the chain of values for depends_on is used. An example might be the clearest way. As it is written, it seems that a value of "." is the final element in the chain, not the first. Also, to ensure that the depends_on chain still works with linked fields, the value of depends_on should be an absolute HDF5 address, not an address relative (such as without any path) to the instance. |
Original reporter: zjttoefs That's all correct. I can provide a link to an example in due course. |
Original reporter: zjttoefs Here's one: |
Thanks, that helps. Next comment shows the structure of that HDF5 file. |
|
There is a conflict between the attribute "offset" as used here and the existing "offset" attribute for fields: The Need to pick a new name here. |
We already use the same attribute names with different meanings depending offset (NX_INT)Rank values of offsets to use for each dimension if the be deprecated and that we say offset (NX_NUMBER) I propose the array stride attribute also be deprecated for the same reason On Thu, Feb 12, 2015 at 5:34 PM, Pete R Jemian notifications@github.com
|
I do agree with Herbert. We need to be more careful in the future and actually check that we do not re-define existing things. I haven't seen file with the "old" offset and that should be rare anyway. The geometry offset is the 99% case. |
Even if we agree to deprecate the offset: Rank values ... attribute, One of the purposes of the NIAC is to avoid such name contention. On 2/12/2015 5:27 PM, Herbert J. Bernstein wrote:
|
Are we also going to rename the newer "signal" attribute? Are all But, if a new name is needed, we could, say, use @base_offset=[...] and On Fri, Feb 13, 2015 at 8:21 AM, Pete R Jemian notifications@github.com
|
Hi all, Am 13.02.2015 um 15:00 schrieb Herbert J. Bernstein <notifications@github.commailto:notifications@github.com>: Are we also going to rename the newer "signal" attribute? Are all But, if a new name is needed, we could, say, use @base_offset=[...] and I think we were on this offset topic in Telcos in last year. My notes say that we decided to rename Best Regards,
On Fri, Feb 13, 2015 at 8:21 AM, Pete R Jemian <notifications@github.commailto:notifications@github.com>
— |
That's the answer then. Thanks.
|
I don't have a particular view on this issue, but I am a little concerned about making changes during unminuted telcos. Shouldn't these decisions then be put to a vote of the NIAC, if only to make sure they are documented? |
see changeset ed22839 |
The change from ed22839 was at line 829: "offset" changed to "data_offset" and a note added at the end of the attribute's documentation. |
You are correct, Ray. Telcos now are minuted, that's the first half of the problem. Whether we want to vote now (online) or delay that until the next NIAC or not vote at all, I'm not sure about. I'd probably go with the middle, but send out for comments. |
Hi, Am 13.02.2015 um 16:41 schrieb Tobias Richter <notifications@github.commailto:notifications@github.com>: You are correct, Ray. Telcos now are minuted, that's the first half of the problem. Whether we want to vote now (online) or delay that until the next NIAC or not vote at all, I'm not sure about. I'd probably go with the middle, but send out for comments. as I remember it, at the Telco we were of the opinion that this does not require a vote. It is a small technical Regards,
— |
Need more understanding of both "vector" and "offset" to make the documentation useful. The example in the GitHub issue is useful but only to a certain point. Not enough to provide clear understanding.
(Note: "depends_on", not "depends") This points to the next transformation in the chain:
and the dependencies chain ends with:
Note the field y_pixel_offset:NX_FLOAT64[2527] and then the transformation fields rotation:NX_FLOAT64[450] and translation:NX_FLOAT64[450]? Why the different array lengths? How is this handled? Might help to show the math as well. |
other places in the manual, check the index for depends_on (since search is not working now) |
assigned to me to add to the documentation |
Original reporter: prjemian
see: http://wiki.nexusformat.org/Main_Page/NXdetector_2012_10
also see #112, #247, #268, and [1168]
used in (amongst others): NXtransformations, NXmx
From Tobias:
depends_on
has crept into a number of places in the documentation now. My short summary:It does exist both as a field for placing components (like NXdetector) and as an attribute to chain together transformations.
A depends_on of "." is the explicit (and required) way of depending on the origin of the coordinate system.
The text was updated successfully, but these errors were encountered: