Re: charset; logical OR on filtering; list of dates

Hi John,

Thank you for your reply, and I am sorry for the late response (I had
no time to work on these nowadays).

> Hi Nagy:
> 
> In message <20210512000436.0a2b1a63@Dell>, Nagy Gabor writes:
> >Now there are 3 questions in my mind about roundup (the last one is a
> >bit more general): 
> >
> >1. The first one is probably a lame question.
> >How can I serve the content of a text/plain file (from a FileClass)
> >with a specified encoding? To be more precise, I want my browser to
> >open http://localhost:8080/tracker/msg4/textfile.txt
> >with "Content-Type: text/plain; charset=UTF-8" header.
> >I do not know how to achieve this using a standalone roundup server
> >(or running roundup-server behind an Apache proxy).  
> 
> Does changing the type property of the file to:
> 
>   text/plain; charset=UTF-8
> 
> result in sending an octet-stream? I think the type even though it's a
> string is sanitized to a list of known types for security. I don't
> remember seeing any other value used other than a plain mime type.
> 
> If my guess above is correct then I don't have an answer other than
> code changes. You could replace Client::serve_file to allow charset.
> I think can be done in interfaces.py. Alternatively add a "charset"
> string property to the FileClass and replace Client::_serve_file
> to emit:
> 
>   self.additional_headers['Content-Type'] = "%s;
> charset=%s"%(mime_type,charset)
> 
> with suitable value assigned to charset. I don't think you can wrap
> _serve_file to modify the Content-Type header while calling the
> original _serve_file to do all the real work. There is a
> self.write(content) call in _serve_file that I think triggers output
> to the browser before _serve_file returns.

Well, I changed the type of txt file to "text/plain; charset=UTF-8", and
"curl -I" reports that the Content-type is "application/octet-stream".
(I gave Admin role to the anonyomous user to let curl download the
file. ;) So I will try to patch the code then. (I hope it will work, I
do not know much about http headers).
 
> >2. Can I do logical OR operation on database filtering, concerning
> >different properties?
> >For example, I want to list issues with "assignedto=15 OR actor=15".
> >Do we have a built-in query for this, or shall I utilize filter_sql,
> >or any other custom solution?  
> 
> I don't know of any generic way to do this. I played with performing
> set operations between queries at a previous employer. The POC used
> two defined searches.  This allowed you to:
> 
>   search1 UNION search2  - this is your OR operation
>   search2 INTERSECT search3 
>   search1 DIFFERENCE search4 
> 
> I only implemented the UNION. To do this I defined a new action:
> 
>   setsearch
> 
> with arguments: set1, set2, op, columns, sort, group
> 
>   @op = UNION (never got around to coding INTERSECTION or DIFFERENCE)
>   @set1 was the designator of a query: query1 query22 etc.
>   @set2 values as set1
> 
>   columns, sort and group were the same as for the regular query
>   interface.
>  
> For UNION and INTERSECTION set1 and set2 are interchangable. For
> DIFFERENCE changing the order of set1 and set2 provides different
> results. This was done before the current query mechanism was
> defined. As a result all of the searches you defined were listed in
> the left sidebar. With the current query management interface, these
> additional queries can be hidden.
> 
> If I was to do it again I might use an RPN style method similar to the
> query mechanism for links/multilinks. So:
> 
>    @q=query1,query2,UNION,query3,DIFFERENCE
> 
> would combine the items from query1 and query2 then from that set
> remove the items in query3. This allows multiple operations among
> queries.
> 
> I also didn't have a method for saving these queries. They were done
> using a template to select the queries and execute the new action.
> This method could allow storing these queries in the query class along
> with the other queries. So they can be displayed in the left hand
> sidebar with the rest.

OK, thanks. I have to dig into the query generation part of the code
(I should especially see how batching works for a query).
 
> >3. This one is a little more general. I know that relational
> >databases are not designed to store lists/arrays. But I have to
> >implement somehow a property (let's call it "important_dates") which
> >stores a LIST of dates. What is the best practice to implement this?
> > 
> 
>  * What do you want to do with the list?
>  * Is it a text blob?
>  * Do you want to search for a value in it?
>  * Will multiple issues link to the same element of the list (e.g.
> date)?
>  * How should it be sorted?
> 
> Basically a list is a mutilink so:
> 
>   issue = IssueClass( ..., important_dates=Multilink(dates), ...)
> 
>   dates=Class(db, date=Date(), order=Integer())
> 
> Have a great day.

I think defining a Class() for dates, maybe combined with rev_multilink
as Ralf suggested, will be fine.

Thanks again,
Gábor

PS. I see that you managed to reproduce my reported [Bug?], so I do no
respond to that email.

> 				-- rouilj
> John Rouillard
> ===========================================================================
> My employers don't acknowledge my existence much less my opinions.
> 
> 
> _______________________________________________
> Roundup-users mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/roundup-users


_______________________________________________
Roundup-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/roundup-users