Re: Question on bounded / unbounded strings

["Dmitry A. Kazakov" <mailbox@dmitry-kazakov.de>]

On 2016-09-24 20:59, John Smith wrote:
> On Saturday, September 24, 2016 at 1:45:14 PM UTC-4, Dmitry A. Kazakov wrote:

>> Searching for a pattern / skipping irrelevant parts of the input do not
>> require that. No need to split anything.
>
> Then how would you accomplish this?

In a loop over the string. There are many algorithms for pattern 
matching and many pattern languages, e.g. wildcards, regular 
expressions, SNOBOL/SPITBOL etc.

BTW splitting does work anyway if search is not anchored, e.g. you need 
to find first occurrence of a pattern.

>> I would use a stream on top of a segmented buffer, e.g.
>>
>> http://www.dmitry-kazakov.de/ada/components.htm#Storage_Streams
>>
>> Or a FIFO if the producer of the content and the SMTP client are run by
>> different tasks.
>>
>> You don't need to build all attachment in the memory in order to be able
>> to send it.
>
> I disagree. If I write out the attachment to disk and then e-mail
> it,  that's even more overhead. Why do this? Create a neat package in memory,
> e-mail it out and then forget about it. Doing this operation in memory
> would be easier and faster. And I don't have to worry about temporary
> files or any other files (as well as deleting the files, testing this
> additional logic, etc.)

No, the idea is that the generator of the attachment sends parts of as 
soon as they are ready. It is a pipeline. The SMTP stack sits on TCP/IP 
stream that sends chunks away.

The stream used as pipeline between the producer and consumer need not 
to be able to hold all attachment. Its internal FIFO buffer can be 
limited, the producer is blocked when the buffer is full and consumer is 
when the buffer is empty.

But my point is that Unbounded_String is almost always a poor substitute 
for more clever and efficient data structures / algorithms. The only 
excuse to use the former is programmer's laziness.

-- 
Regards,
Dmitry A. Kazakov
http://www.dmitry-kazakov.de