On Thu, Dec 23, 2010 at 1:52 AM, Jeremy Goecks <jeremy.goecks@emory.edu> wrote:
That makes sense, but I guess somewhere before the Javascipt runs the start value is being set to the first entry in the list.
Yes, I think this is the case; there is also an 'accept_default' attribute that can be added to select parameters.
Good idea, I tried adding accept_default="false", but it doesn't seem to do anything (the first list entry is still picked, a bug regardless of the search feature?). I also tried combining this with force_select="true", again no change. BTW, having the default value for this and other param attributes as another column in the table on the wiki would be a big help: https://bitbucket.org/galaxy/galaxy-central/wiki/ToolConfigSyntax
I've tried adding value="" and value="?" to the <param type="select" ...> tag to no avail.
Hmm. For genome/dbkey selection, the first value is often '?'; for datatype selection, the first value is often ''. Hence, the reason for those two values being special.
I can insert a dummy entry at the start of my loc file like that, and then I do see the "Click to Search or Select" text. However, the drop down list includes the dummy entry - so this isn't a very good solution. I'd need to add a validator to block the dummy entry or something, but it would still be distracting.
Maybe I'm missing something. Should all browsers behave the same?
Yes, all browsers should behave the same, modulo some IE quirks.
Can you provide the XML for your param and I can take a look?
Easy, I'm currently playing with blastn (or any of the BLAST+ tools using nucleotides databases since I have lots of them. The blastp wrapper would have the same issues except currently I have less than 20 protein databases setup. See: https://bitbucket.org/galaxy/galaxy-central/src/default/tools/ncbi_blast_plu... Currently my best workaround so far is to add a help attribute, "When there are lots of entries, type here to search" (this needs to be worded carefully as when there are only a few databases setup you can't search). Thanks, Peter P.S. I will still be checking email, but our Institute closes today for the Christmas break, so I probably won't be able to try and code changes immediately.