Galaxy lists empty file for composite datatypes

classic Classic list List threaded Threaded
1 message Options
| Threaded
Open this post in threaded view

Galaxy lists empty file for composite datatypes

Peter Cock
Hello all,

In replying to someone having trouble with the BLAST+ tools, I
found a new bug (and was reminded of an old bug I'd not yet reported)
in the handling of composite datatypes without a primary file (e.g.
BLAST databases).

Consider the screenshot, where the history text for my BLAST database
says "empty" (above the line saying "format: blastdbn") and the
"full path" (highlighted) shown via the "information" icon points at an
empty file, in this case:

$ ls /mnt/galaxy/galaxy-dist/database/files/009/dataset_9085.dat -l
-rw-r--r-- 1 galaxy galaxy 0 May 17 14:40

The real data is here:

$ ls /mnt/galaxy/galaxy-dist/database/files/009/dataset_9085_files/
blastdb.nhd  blastdb.nhi  blastdb.nhr  blastdb.nin  blastdb.nog
blastdb.nsd  blastdb.nsi  blastdb.nsq

I am guessing the code assumes a composite datatypes like HTML
with a primary file, where it makes some sense to measure the
size of the primary file (although this omits the child files like
images for an HTML file) and use its location on disk.

In the case of basic composite datatypes like BLAST databases,
the "full path" should be the folder name, and the size should
count all the child files within that folder (which is NOT empty).

Looking at the code,

$ grep "Full Path:" */*
templates/show_params.mako:            <tr><td>Full
Path:</td><td>${hda.file_name | h}</td></tr>

This suggests that the hda.file_name attribute might be assigned
the folder name in this case? Or might that have repercussions?



Please keep all replies on the list by using "reply all"
in your mail client.  To manage your subscriptions to this
and other Galaxy lists, please use the interface at:

To search Galaxy mailing lists use the unified search at:

Screen Shot 2013-05-17 at 14.46.55.png (201K) Download Attachment