Arbitrary PCD changes

FvM · Joined: 27 Aug 2008 Posts: 2337 Location: Germany

I just noticed, that in PCD V4.106 previously existing interrupt definitions have been erroneously removed from the
device file, in this case INT_DMAERR for all PIC24FJ chips. It's rather annoying, that I apparently have to expect
compilation failure of existing applications with each new PCD version.

I would expect, that these kind of artefacts are avoided by basic quality control.

dezso · Joined: 04 Mar 2010 Posts: 102

This is reason #27 why I hesitate to spend $600 on this.

FvM · Joined: 27 Aug 2008 Posts: 2337 Location: Germany

CCS support clarified, that the said INT_DMAERR isn't available in PIC24FJ hardware. Unfortunately,
it had been included in the PCD valid interrupts documentation up to V4.105.

So in this case, CCS didn't introduce a new bug, but intended to remove an existing specification error (that
did no harm yet). As usual, it's undocumented in the "recent changes" list.

FvM · Joined: 27 Aug 2008 Posts: 2337 Location: Germany

I was just aware, that the code for spi_read() built in function in PCD, that has been working for two years has been
replaced by erroneous code in V4.109. It's causing a watchdog reset in my application.

dyeatman · Joined: 06 Sep 2003 Posts: 1933 Location: Norman, OK

It does need to be said that 4.099 is the last version CCS considers a stable
release (shown on the upper right side of the download page). All releases
after that are considered beta. I keep two copies installed. One is the
stable release (version 4.099) I use for production and the other is the latest
release (4.109 in this case) for testing.

I have been hoping for years they would change the wording on their web
site to reflect this....maybe one day!
_________________
Google and Forum Search are some of your best tools!!!!

FvM · Joined: 27 Aug 2008 Posts: 2337 Location: Germany

I didn't yet notice a remark on the update page about "stable" or beta versions. In my opinion, V4.099 has been kept because many changes have been officially introduced starting with V4.100. Does this mean, that V4.099 is more stable than later versions? I fear, it isn't. A large number of bugs has been continuously fixed with new releases, some are still continued, a few have been newly introduced, as in the present case.

But in any case, you should keep a compiler version, that has been used for a production release.

dyeatman · Joined: 06 Sep 2003 Posts: 1933 Location: Norman, OK

Historically, since back in version 3, the version on the right of the Download
page has been the latest "full and stable" release and the version on the left
is considered beta. I have been a CCS user since 2003 and I believe that is
still the case. Like I said earlier, it would be nice if CCS was up front about
this.

PCM and TTelmah, please correct me if I am wrong, is this the way you understand it?
_________________
Google and Forum Search are some of your best tools!!!!

Ttelmah · Joined: 11 Mar 2010 Posts: 19504

Basically yes.
Except when they are doing major updates, when the 'new' version is more like an 'alpha'....

The thing that makes it really hard to know what is likely to give problems, is the poor documentation of the changes. They basically report 'one thing' changed with a release, when in reallity most have several dozen parts tweaked.

My own approach is always to document in the source code, what compiler was used with it, and when I archive sources, archive the compiler used with it. When I try a new compiler, I keep the old, till I have verified that the code I am working on, works with the new version. Once this is 'true', update the note in the source, to say this. The note is always 'cumulative'. So (for example):

FvM · Joined: 27 Aug 2008 Posts: 2337 Location: Germany

spi_read() is O.K. again with PCD V4.110, also spi_write() seems to work finally as expected.

But why this #FUSE name has been changed?