timj / aandc-fits Goto Github PK
View Code? Open in Web Editor NEWPaper on the FITS format for Astronomy and Computing journal
Paper on the FITS format for Astronomy and Computing journal
Hi, section 2.4.4. is interesting, but I am worried that we dont have enough support here for why compression in FITS is unsuitable. The FITS community largely feel that this is one of the details that FITS gets 'right'. From my reading of this section, what I think is being advocated here is not that tiled (rice) compression is bad, but rather that its only implemented in a convention, and other parsers/readers wont understand it, correct?
If thats the case, then what the underlying problem is that our readers lack an API which is 'extensible' via a shared plugin mechanism (to point out one possible solution). Do I have that right? Assigning to you Slava since I think you created this section (if not, and you have no opinions, please assign back to me). -brian
The relative placement (and implied relative importance) of Sec 2.1.3, "No support for alternative encodings", and Sec 2.3.1, "Constrained metadata representation", looks inverted to me. At the end of 2.1.3 it is stated that expressing FITS keywords more broadly is less critical than 7-bit ASCII. Most people are much, much more bothered by the 8-character limit on keyword names than by lack of Unicode.
In terms of relative importance, I propose 2.1.3 be subsumed into 2.3.1.
At minimum, the end of 2.1.3 should say that the important issue of keyword naming is addressed later.
Time line for submission will be sent round by @brianthomas
commit 75aed13 broke the references when compiling the paper using pdflatex.
@GBruceB has volunteered to have a look at the provenance section.
Section 2.1 seems mistitled at "Poor support for information interchange". Isn't what is discussed the hierarchies or relationships between data elements?
Herschel provides a nice example of implementing hierarchies. They are done as FITS files. For Herschel, the Observation Context is the container for Calibration, Pointing, Raw Data, Processed Data, etc. and would make a nice diagram to illustrate the point.
(I brought this up for the ADASS paper but had trouble finding a reference. However if Wikipedia is used as a reference for the definition of 'data model', can't we use the online Herschel documentation as a similar reference?)
When reading Sec 2.1.2, 5th paragraph about having multiple data models, it occurred to me that a nice practical example is distortion metadata in images from the Palomar Transient Factory (PTF). That project decided to incorporate both "SIP" and "TPV" keywords in their headers, to make sure that both SExtractor and other tools could handle their data. This leads to having to make rules of what to do when both data models are present.
For a reference, see http://proceedings.spiedigitallibrary.org/proceeding.aspx?articleid=1363103
also available at at http://web.ipac.caltech.edu/staff/shupe/reprints/SIP_to_PV_SPIE2012.pdf
Section 3.1 in the latest draft references "AstroPy", but we spell it "Astropy" (with a lower-case "p").
This is my short list of issues (other than those in #39) with the text:
model1:TEMP
and model2:TEMP
(i.e., using \texttt{})And that's all on my part.
I was wondering about using the GitHub issues facility to make questions on things in the article that might need discussion, before any kind of actual commit to the paper. If you (specially @timj as the "host") agree, I think it's easier than just having the conversation on email, as it makes for more "on topic" conversation. These issues would work similarly to how we worked with the Google Doc notes.
Hi Slava,
I think you added this content (correct me if thats wrong). Could you please add in the references to the bib file? They appear to have gone missing.
I think that section 2.1.3 (No support for alternative encodings) could be move to section 2.3 on inflexibility in information representation because it is more related with the issues discussed there whereas section 2.1 deals mostly with practical aspects of data transfer.
CLASS and Miriad uvdata and image formats that are used in radio astronomy could be introduced in one of the paragraphs in the introduction.
There are some comments I gave on the draft of the paper @timj sent out to the astropy list:
In this short section, mention is made that LOFAR has been driven in part to use HDF5 solution instead of FITS because of the need for parallel read/write. Can a few sentences be written to describe the solution? Im guessing that as an ex-LOFAR person Anastasia, that you can provide this detail? If you are unable to do so, just assign back to me and I'll try to grok it from your paper. -brian
Otherwise I am very happy with the text. Thanks for the good work!
As discussed in #12, it would be good to add a section discussing FITS's poor support for missing values (distinct from NaN). Not only does TNULL not work for float columns, for integer columns it requires sacrificing a value to be the 'NULL' value. Ideally a format should fully support masking of values.
This week is a little crazy for me, so I won't have time to contribute text to it, but if no one else has time and the paper is still being prepared after the new year, then I can work on something then.
After the latest edits, the footnote for once fits, forever fits is split between pages, and this causes a \pdfendlink ended up in different nesting level than \pdfstartlink
error.
This error and solution are characterised here 1, so this should be solved after other content is added.
The workaround, as suggested in 1, is to add draft to the \usepackage{hyperref}
part, but the problem must be solved before submission.
Section 2.1 3rd paragraph lists several examples of hierarchical formats (I think). I find I'm familiar with none of them so this part leaves me befuddled. Would it be possible to expand upon the benefits of e.g. the Starlink HDS format? Why do arbitrary hierarchies lead to chaos? Can an example be provided?
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.