cmake eclass fix

07-07-2011, 02:42 PM

Hi all,

Can someone replace the link the cmake eclass that says ....

if use prefix; then

to

if use prefix && ! EPREFIX=; then

or similar.

Thanks,

Alan.

08-14-2011, 10:40 AM

On 07-07-2011 15:42:40 +0100, Alan Hourihane wrote:
> Hi all,
>
> Can someone replace the link the cmake eclass that says ....
>
> if use prefix; then
>
> to
>
> if use prefix && ! EPREFIX=; then

Why is that necessary? Or is it MiNT that doesn't like the RPATH stuff?

--
Fabian Groffen
Gentoo on a different level

08-14-2011, 06:05 PM

On 08/14/11 11:40, Fabian Groffen wrote:
> On 07-07-2011 15:42:40 +0100, Alan Hourihane wrote:
>> Hi all,
>>
>> Can someone replace the link the cmake eclass that says ....
>>
>> if use prefix; then
>>
>> to
>>
>> if use prefix && ! EPREFIX=; then
> Why is that necessary? Or is it MiNT that doesn't like the RPATH stuff?
>
Yes, it's the RPATH stuff that messes things up when EPREFIX=/

Alan.

08-14-2011, 06:06 PM

2011-8-15 ä¸Šåˆ2:05 æ–¼ "Alan Hourihane" <alanh@fairlite.co.uk> å¯«é“ï¼š> On 08/14/11 11:40, Fabian Groffen wrote:
>> On 07-07-2011 15:42:40 +0100, Alan Hourihane wrote:

>>> Hi all,
>>>
>>> Can someone replace the link the cmake eclass that says ....
>>>
>>> if use prefix; then
>>>
>>> to
>>>
>>> if use prefix && ! EPREFIX=; then

>> Why is that necessary? Or is it MiNT that doesn't like the RPATH stuff?
>>
> Yes, it's the RPATH stuff that messes things up when EPREFIX=/
>
> Alan.
>

10-04-2011, 10:21 AM

On 08/14/11 19:05, Alan Hourihane wrote:
> On 08/14/11 11:40, Fabian Groffen wrote:
>> On 07-07-2011 15:42:40 +0100, Alan Hourihane wrote:
>>> Hi all,
>>>
>>> Can someone replace the link the cmake eclass that says ....
>>>
>>> if use prefix; then
>>>
>>> to
>>>
>>> if use prefix && ! EPREFIX=; then
>> Why is that necessary? Or is it MiNT that doesn't like the RPATH stuff?
>>
> Yes, it's the RPATH stuff that messes things up when EPREFIX=/
>

Any better fix for this ?

Alan.

10-04-2011, 10:24 AM

On 10/04/11 11:21, Alan Hourihane wrote:
> On 08/14/11 19:05, Alan Hourihane wrote:
>> On 08/14/11 11:40, Fabian Groffen wrote:
>>> On 07-07-2011 15:42:40 +0100, Alan Hourihane wrote:
>>>> Hi all,
>>>>
>>>> Can someone replace the link the cmake eclass that says ....
>>>>
>>>> if use prefix; then
>>>>
>>>> to
>>>>
>>>> if use prefix && ! EPREFIX=; then
>>> Why is that necessary? Or is it MiNT that doesn't like the RPATH stuff?
>>>
>> Yes, it's the RPATH stuff that messes things up when EPREFIX=/
>>
> Any better fix for this ?
>
> Alan.

if use prefix && "${EPREFIX}" != ""; then

The above is better. Can this be committed ?

Thanks,

Alan.

10-05-2011, 07:01 PM

On 10/04/11 11:24, Alan Hourihane wrote:
> On 10/04/11 11:21, Alan Hourihane wrote:
>> On 08/14/11 19:05, Alan Hourihane wrote:
>>> On 08/14/11 11:40, Fabian Groffen wrote:
>>>> On 07-07-2011 15:42:40 +0100, Alan Hourihane wrote:
>>>>> Hi all,
>>>>>
>>>>> Can someone replace the link the cmake eclass that says ....
>>>>>
>>>>> if use prefix; then
>>>>>
>>>>> to
>>>>>
>>>>> if use prefix && ! EPREFIX=; then
>>>> Why is that necessary? Or is it MiNT that doesn't like the RPATH stuff?
>>>>
>>> Yes, it's the RPATH stuff that messes things up when EPREFIX=/
>>>
>> Any better fix for this ?
>>
>> Alan.
> if use prefix && "${EPREFIX}" != ""; then
>
> The above is better. Can this be committed ?
>

Actually this works for me.....

if use prefix && ! ${EPREFIX}; then

Alan.

10-05-2011, 09:39 PM

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA512

On 10/05/2011 03:01 PM, Alan Hourihane wrote:
> On 10/04/11 11:24, Alan Hourihane wrote:
>> On 10/04/11 11:21, Alan Hourihane wrote:
>>> On 08/14/11 19:05, Alan Hourihane wrote:
>>>> On 08/14/11 11:40, Fabian Groffen wrote:
>>>>> On 07-07-2011 15:42:40 +0100, Alan Hourihane wrote:
>>>>>> Hi all,
>>>>>>
>>>>>> Can someone replace the link the cmake eclass that says
>>>>>> ....
>>>>>>
>>>>>> if use prefix; then
>>>>>>
>>>>>> to
>>>>>>
>>>>>> if use prefix && ! EPREFIX=; then
>>>>> Why is that necessary? Or is it MiNT that doesn't like the
>>>>> RPATH stuff?
>>>>>
>>>> Yes, it's the RPATH stuff that messes things up when
>>>> EPREFIX=/
>>>>
>>> Any better fix for this ?
>>>
>>> Alan.
>> if use prefix && "${EPREFIX}" != ""; then
>>
>> The above is better. Can this be committed ?
>>
>
> Actually this works for me.....
>
> if use prefix && ! ${EPREFIX}; then
>
> Alan.
>

I hope you meant:

if use prefix && ! [[ ${EPREFIX} ]]; then

- --
Jonathan Callen
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.18 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iQIcBAEBCgAGBQJOjM54AAoJELHSF2kinlg4qkwQAKajXUgpGO trFFEJIl9Nw0kP
Sel9cTC6OLf0H+Ct2yeVKHLFmQhWD/56itWXlS/XzFvQk18d4+KdxRd/99AvHBEb
Rc2Rz44wGxkYf/CTnshiWARX5uePUNGTRPbe75gOf6lssMyqtbvvROKepPpI1ZfJ
kJFji8Zag2V+pO8ulvXp5BQMrN83rPzJlixqkTvhH5YN+i+1rO lTkfyA9KdKjFeP
7HChK00M3u+Ni1DsJJmvjUXCkEQL+FqN9zXQwG7WLe+QC2DiG9 n9vrAuuvQUzQYZ
qMed9AvYLJ5EaVj3YLlNwBhNDdfE4DO0ndyjGUxsWiDsbcXsm0 N0JiyIN62bJIUD
RyqusX5r8zhtbImO6oVZ3UvUA1t+48Cxi5yvoKN1H5UejT5nCR 85fWdZ2P+QtAb6
LhDGj7ODC3KYyq1MWdPcQ6LmNDEjryzWkrUL6SfKvAptrqkOrg 3dRiLNPmy3GiII
8PfyA1m8aiIP0Cyv8p94jnxt7rj34qUobEGbIT3NcUdeu6pNK/ozspYeA5bX7jXB
Xo8VwC6BRlsMic3u500hwUX7kbcLbvH1cyX2wj/RdtJsvVB8iQyDxoaoUo6yaiwI
JIiOQfdUfR3oA/4Hcl6ek8qz1BDYDdLiXS7esLxmMwCTh1IpJngDg40Mauw3waqZ
h6CI8SlyxavFCLI4M1Mq
=000s
-----END PGP SIGNATURE-----

10-05-2011, 10:33 PM

On 10/05/11 22:39, Jonathan Callen wrote:
> On 10/05/2011 03:01 PM, Alan Hourihane wrote:
> > On 10/04/11 11:24, Alan Hourihane wrote:
> >> On 10/04/11 11:21, Alan Hourihane wrote:
> >>> On 08/14/11 19:05, Alan Hourihane wrote:
> >>>> On 08/14/11 11:40, Fabian Groffen wrote:
> >>>>> On 07-07-2011 15:42:40 +0100, Alan Hourihane wrote:
> >>>>>> Hi all,
> >>>>>>
> >>>>>> Can someone replace the link the cmake eclass that says
> >>>>>> ....
> >>>>>>
> >>>>>> if use prefix; then
> >>>>>>
> >>>>>> to
> >>>>>>
> >>>>>> if use prefix && ! EPREFIX=; then
> >>>>> Why is that necessary? Or is it MiNT that doesn't like the
> >>>>> RPATH stuff?
> >>>>>
> >>>> Yes, it's the RPATH stuff that messes things up when
> >>>> EPREFIX=/
> >>>>
> >>> Any better fix for this ?
> >>>
> >>> Alan.
> >> if use prefix && "${EPREFIX}" != ""; then
> >>
> >> The above is better. Can this be committed ?
> >>
>
> > Actually this works for me.....
>
> > if use prefix && ! ${EPREFIX}; then
>
> > Alan.
>
>
> I hope you meant:
>
> if use prefix && ! [[ ${EPREFIX} ]]; then
>
I'd be fine with that :-)

Alan.

10-06-2011, 09:15 AM

Hi Alan,

On 10/06/2011 12:33 AM, Alan Hourihane wrote:
> On 10/05/11 22:39, Jonathan Callen wrote:
>> On 10/05/2011 03:01 PM, Alan Hourihane wrote:
>>> On 10/04/11 11:24, Alan Hourihane wrote:
>>>> On 10/04/11 11:21, Alan Hourihane wrote:
>>>>> On 08/14/11 19:05, Alan Hourihane wrote:
>>>>>> On 08/14/11 11:40, Fabian Groffen wrote:
>>>>>>> On 07-07-2011 15:42:40 +0100, Alan Hourihane wrote:
>>>>>>>> Hi all,
>>>>>>>>
>>>>>>>> Can someone replace the link the cmake eclass that says
>>>>>>>> ....
>>>>>>>>
>>>>>>>> if use prefix; then
>>>>>>>>
>>>>>>>> to
>>>>>>>>
>>>>>>>> if use prefix && ! EPREFIX=; then
>>>>>
>>>> if use prefix && "${EPREFIX}" != ""; then
>>>>
>>>> The above is better. Can this be committed ?
>>>>
>>
>>> Actually this works for me.....
>>
>>> if use prefix && ! ${EPREFIX}; then
>>
>> I hope you meant:
>>
>> if use prefix && ! [[ ${EPREFIX} ]]; then

Reading the different variants (without reading the cmake.eclass),
these variants do give different results.

Probably, you do want to fire that 'if' when "EPREFIX is not empty", right?

Actually, the last one (as reported in [1]) does fire "when EPREFIX is empty".

So you probably want:

- if use prefix; then
+ if use prefix && [[ ${EPREFIX} ]]; then

So - have you actually tried your reported variant?

<pointless optimization>
Given that EPREFIX is set with 'use prefix' only, it could also read:

- if use prefix; then
+ if [[ ${EPREFIX} ]]; then
</pointless optimization>

OTOH: Do we support non-empty EPREFIX on FreeMiNT?

[1] https://bugs.gentoo.org/show_bug.cgi?id=385839

/haubi/