Board index » cppbuilder » BCB why not compile as thread?

BCB why not compile as thread?

Hello,

when I go to start and BCB begins to compile and link, everythig ist
blocked. Is it possible to configure it that maybe I can browse Helpfiles or
my Sources? Or is this planned for future Versions of BCB? I guess this
could not be so difficult to implement, other compilers have this for a long
time.

Harald

 

Re:BCB why not compile as thread?


: "Harald Plontke" <harald.plon...@cruise.de> wrote:

Quote
>when I go to start and BCB begins to compile and link, everythig ist
>blocked. Is it possible to configure it that maybe I can browse Helpfiles or
>my Sources? Or

As of BCB4 this is not possible - you could try going to
http://www.bcbdev.com/ and locate Harold Howe's command line compiler
wrapper...

Quote
>I guess this could not be so difficult to implement,

This is what you guess :-)

--
Stefan Hoffmeister (TeamB)     http://www.econos.de/
Please do apply judgement when sending email.

Re:BCB why not compile as thread?


On Tue, 21 Sep 1999 09:41:25 +0200, "Harald Plontke"

Quote
<harald.plon...@cruise.de> wrote:
>when I go to start and BCB begins to compile and link, everythig ist
>blocked. Is it possible to configure it that maybe I can browse Helpfiles or
>my Sources?

Nothing's stopping you from browsing through help files - just have
them open beforehand. As for the code, if you made changes during
compilation how would the de{*word*81} deal with it? Undo your changes?

Re:BCB why not compile as thread?


Quote
Michael Warner wrote:
> Nothing's stopping you from browsing through help files - just have
> them open beforehand. As for the code, if you made changes during
> compilation how would the de{*word*81} deal with it? Undo your changes?

Out of curiosity, does anyone know why BCB doesn't compile in a separate
thread?  Surely its not just because the user might modify the source
code during compilation.  I would gladly obey an arbitrary commandment
such as: "Thou shalt not modify source code during compilation" if it
meant that I could still perform some other very important tasks (such
as minimizing the IDE or reviewing code that I've already written).

If the only problem with modifying code while compiling is that the
de{*word*81} wouldn't know what to do, I could live with another arbitrary
commandment imposed upon the user: "If you modify code during
compilation, all bets are off when using the de{*word*81}."  But this is
what I would expect.

I know that allowing the user to mess around with things during
compilation is a scary proposition and could certainly cause some
problems, and I know that imposing rules upon the user that must be
obeyed or the software will crash is not a good idea in software design,
but I think we could live with a few rules if it meant that we wouldn't
have to take a fif{*word*249} minute coffee break every time we built our
project.

I know that in the future Builder will probably be able to compile in
the background, but in the mean time I think I would feel a lot better
about Borland (and probably grumble less) if I heard some really good
reasons (other than "we can't trust the user not to change something")
that it isn't being done.

Jeff Wilhite

Re:BCB why not compile as thread?


Quote
> I know that allowing the user to mess around with things during
> compilation is a scary proposition and could certainly cause some
> problems, and I know that imposing rules upon the user that must be
> obeyed or the software will crash is not a good idea in software design,
> but I think we could live with a few rules if it meant that we wouldn't
> have to take a fif{*word*249} minute coffee break every time we built our
> project.

> I know that in the future Builder will probably be able to compile in
> the background, but in the mean time I think I would feel a lot better
> about Borland (and probably grumble less) if I heard some really good
> reasons (other than "we can't trust the user not to change something")
> that it isn't being done.

Some felt that it was (and is) a mostly useless feature. Many of us have
not forgotten how useless it was to have in BC 5. Still you are correct
you will likely see background compilation in the next major release due
to customer demand. The rules you are speaking of will likely be
enforced by the IDE making it much more likely that people will follow
them.

As for really good reasons as to why it was not done in the past: It
adds a great deal of complexity to the product for what some may
consider little gain. We must now devise a real code mechanism to
prevent a great deal of what was prevented in single thread operation.

To be on the safe side we really have to tag everything read-only. This
allows you to minimize the IDE, read help, and view source code and
perhaps forms. At the same time the quality of these actions is reduced
as a large chunk of your machine is now both compiling and allowing you
to perform other tasks. Multitasking only works well when there is idle
time. Given the nature of compiling and linking generally little
"natural" idle time exists and so idle time is emulated by running the
compilation in a thread with a lower priority. The end result is
generally slower build times with only "passable" usability during the
build.

--
Anduin Withers
Borland

[ Guidelines for posting to this group can be found here:  ]
[    http://www.borland.com/newsgroups/guide.html          ]
[ Note: posts to servers other than forums.borland.com     ]
[ will not be seen by most newsgroup users.                ]

Re:BCB why not compile as thread?


Quote
Jeff Wilhite wrote:

> Michael Warner wrote:
> > Nothing's stopping you from browsing through help files - just have
> > them open beforehand. As for the code, if you made changes during
> > compilation how would the de{*word*81} deal with it? Undo your changes?

> Out of curiosity, does anyone know why BCB doesn't compile in a separate
> thread?

It does if you use the command line compiler. Compiles as a whole
separate program, in fact.

Of course like any self-respecting C++ compiler, it pegs your CPU so
that you can't do anything else anyway.

Lin
linsh...@gte.net

Re:BCB why not compile as thread?


I'm all for compilation in its own thread! You can, however, currently open
a second copy of Borland and do your thing... though it's likely Borland
aren't going to encourage this - one must take care when doing this. I used
to do this but now I find using 2 PCs works better. Regarding Help files,
you can open them during compilation off your Start menu.
Quote

> Some felt that it was (and is) a mostly useless feature. Many of us have
> not forgotten how useless it was to have in BC 5. Still you are correct
> you will likely see background compilation in the next major release due
> to customer demand. The rules you are speaking of will likely be
> enforced by the IDE making it much more likely that people will follow
> them.

> As for really good reasons as to why it was not done in the past: It
> adds a great deal of complexity to the product for what some may
> consider little gain. We must now devise a real code mechanism to
> prevent a great deal of what was prevented in single thread operation.

> To be on the safe side we really have to tag everything read-only. This
> allows you to minimize the IDE, read help, and view source code and
> perhaps forms. At the same time the quality of these actions is reduced
> as a large chunk of your machine is now both compiling and allowing you
> to perform other tasks. Multitasking only works well when there is idle
> time. Given the nature of compiling and linking generally little
> "natural" idle time exists and so idle time is emulated by running the
> compilation in a thread with a lower priority. The end result is
> generally slower build times with only "passable" usability during the
> build.

> --
> Anduin Withers
> Borland

> [ Guidelines for posting to this group can be found here:  ]
> [    http://www.borland.com/newsgroups/guide.html          ]
> [ Note: posts to servers other than forums.borland.com     ]
> [ will not be seen by most newsgroup users.                ]

Re:BCB why not compile as thread?


Some applications are considerably difficult to make multi-threaded if it
wasn't designed into it from the beginning.  There are locking issues,
disabling/enabling menus, and lots of code changes all over the place that
would probably have to occur to do that.  Essentially, it requries "modes"
within the IDE, which if they were not built in (which I'm assuming they're
not) then they must be retro-fitted.  Considering the size of the project,
it is not surprising to me that it has not been done.  The benefits are
minimal, and the effort is large.  However, because Microsoft does it people
seem to think that it's something required.  Personally, I don't care one
way or the other about it, because I can easily browse my code in Emacs, and
read help in a seperate process, even while it is compiling.

Also, compile times are going to be slower due to all the other stuff going
on and the compiler having to yield CPU cycles when you want to open a menu,
etc.

Chris (TeamB)

Quote
Jeff Wilhite wrote in message <37E7B329.C2494...@acm.org>...
>I know that in the future Builder will probably be able to compile in
>the background, but in the mean time I think I would feel a lot better
>about Borland (and probably grumble less) if I heard some really good
>reasons (other than "we can't trust the user not to change something")
>that it isn't being done.

Re:BCB why not compile as thread?


Quote
Chris Uzdavinis (TeamB) wrote:
> it is not surprising to me that it has not been done.  The benefits are
> minimal, and the effort is large.  However, because Microsoft does it people
> seem to think that it's something required.  Personally, I don't care one

Whether the benefits are minimal or not is a matter of opinion and the sweeping
statement about what people think because of Microsoft doesn't apply to me.

I see it as there being 2 alternative philosophies.  You can let people do
stupid things so they can also do the useful things they want to do or you can
stop people doing useful things to prevent them doing stupid things.
Personally, I prefer the first approach because I don't like having what I can
do arbitrarily curtailed.  If I do something stupid as a consequence, that's MY
problem.

Quote
> Also, compile times are going to be slower due to all the other stuff going
> on and the compiler having to yield CPU cycles when you want to open a menu,
> etc.

Again, MY problem if I do something which guzzles CPU time.  I don't want to be
prevented from doing this because just I might actually use it.  If I want
compilation speed, I'll avoid doing CPU intensive things at the same time.

Re:BCB why not compile as thread?


On Wed, 22 Sep 1999 09:53:38 +0930, Radar User <swaw...@atrad.com.au>
wrote:

Quote
>I see it as there being 2 alternative philosophies.  You can let people do
>stupid things so they can also do the useful things they want to do or you can
>stop people doing useful things to prevent them doing stupid things.

In other words, Inprise decided not to do it because they don't think
they can trust you :-)

Quote
>Personally, I prefer the first approach because I don't like having what I can
>do arbitrarily curtailed.  If I do something stupid as a consequence, that's MY
>problem.

But the deluge of complaints which results from allowing people to
blow off their feet isn't your problem, is it?

Re:BCB why not compile as thread?


Quote
Anduin Withers (Borland) wrote in message <37E7C34C.5860E...@inprise.com>...
>Some felt that it was (and is) a mostly useless feature. Many of us have
>not forgotten how useless it was to have in BC 5. Still you are correct
>you will likely see background compilation in the next major release due
>to customer demand. The rules you are speaking of will likely be
>enforced by the IDE making it much more likely that people will follow
>them.

Dear Anduin,

I am a longish time user of BCB and I am very positive about it.  The errors
in the version IDE are, however, a shame to say the least.  I know they do
not affect the end user's experience of my programs, but these problems are
in my face all day every day, because programming is what I do for a living
and the tool I am using at the moment is BCB version 4.

I used to have the theory that hardware manufacturers put their least
competent people on designing the boxes for their computers and that
software companies put the intellectually challenged on writing installation
scripts.  I am now beginning to wonder who gets to write IDEs these days,
having experienced constant annoyances from silly errors from Microsoft,
Watcom and now Borland.  Come on, guys, an IDE is not exactly rocket science
for goodness sake.  Have a little self respect and do the job right.

With all due respect the whoever makes the decisions as to where development
capacity gets spent, I submit that fixing all the really stupid and annoying
bugs in the current IDE ASAP would do more to win the hearts and minds of
your customers than any amount of bells and whistles added to an already
unstable product.

Don't get me wrong, I think BCB is an excellent tool, I value your excellent
compiler and I truly appreciate the lack of errors in the VCL.  If you can
get that right, you must be able to solve a silly thing like an IDE.  As I
said, the IDE errors are in my face all day and I really feel annoyed that
your Quality department has let such a slipshod product out of the door.  I
work for a small software company who could not stay in business 5 minutes
if it released versions of our product in this state.  This quality level
would be a firing offence here.

Perhaps you might be inclined to fee this back to your management.  They
might listen to you more than they appear to be listening to us at the
moment.

Re:BCB why not compile as thread?


: "Guy H. Broadfoot" <gu...@acm.org> wrote:

Quote
>I submit that fixing all the really stupid and annoying
>bugs in the current IDE ASAP

Could you please go to

  http://www.borland.com/devsupport/bugs/bug_reports.html

and submit a bug report on each bug?

Only by doing that do you make sure that the bugs are *known*. And, as you
know, only *known* bugs can possibly be fixed.

Thanks!

--
Stefan Hoffmeister (TeamB)     http://www.econos.de/
Please do apply judgement when sending email.

Re:BCB why not compile as thread?


Quote
Michael Warner wrote:
> >Personally, I prefer the first approach because I don't like having what I can
> >do arbitrarily curtailed.  If I do something stupid as a consequence, that's MY
> >problem.

> But the deluge of complaints which results from allowing people to
> blow off their feet isn't your problem, is it?

Good point.  But if I do something stupid, when I find out I learn not to do it
again (=> 1 complaint) and I can answer other people's questions about it in this
newsgroup.  If there's something the software won't let me do, I don't learn how to
get around it the first time I run into it, I remain unhappy with it for ever more
(=> many complaints) and when someone else complains in the newsgroup, I can only
agree with them.

Re:BCB why not compile as thread?


I know the popular belief is that nothing anyone says on the forums ever
makes it beyond TeamB. This is, at least in part, inaccurate. Often a
thread will be forwarded to higher ups. I don't think anyone on TeamB or
any Borland employee on these groups feels all information should be
filtered through them.

Still imagine this: I'm in a meeting and the IDE becomes the topic. The
most I can possibly forward of your post is that "someone of the groups
says the IDE sucks". I'd probably choose to leave off the bit about
believing we are staffing the IDE development team with idiots as such
statements tend to make people (especially those on the IDE team) more
hostile and less receptive to your ideas.

If you have valid bugs you want addressed the best way to get that
message across to us is to post here so we can all discuss the problem
and then submit a bug report.

This allows me to be in the same meeting and say "someone on the groups
says the IDE sucks". This invariably will lead into a discussion of
exactly what sucks and I will then be able to say that X, Y, and Z
really suck but W only sucks on Tuesdays when generate linker state
files is enabled.

--
Anduin Withers
Borland

[ Guidelines for posting to this group can be found here:  ]
[    http://www.borland.com/newsgroups/guide.html          ]
[ Note: posts to servers other than forums.borland.com     ]
[ will not be seen by most newsgroup users.                ]

Re:BCB why not compile as thread?


Dear Stefan,

I will certainly do this, although no-one who reads the news groups can
claim to be unaware of the general quality of the IDE at the moment,
including people from Borland.

Regards
Guy

Quote
Stefan Hoffmeister (TeamB) wrote in message

<=F3pN4Thnks=uWqVVc60tnt4I...@4ax.com>...
Quote
>: "Guy H. Broadfoot" <gu...@acm.org> wrote:

>>I submit that fixing all the really stupid and annoying
>>bugs in the current IDE ASAP

>Could you please go to

>  http://www.borland.com/devsupport/bugs/bug_reports.html

>and submit a bug report on each bug?

>Only by doing that do you make sure that the bugs are *known*. And, as you
>know, only *known* bugs can possibly be fixed.

>Thanks!

>--
>Stefan Hoffmeister (TeamB)     http://www.econos.de/
>Please do apply judgement when sending email.

Go to page: [1] [2]

Other Threads