Processed: HDF 4.3.1 uploaded to unstable

[Antonio Valentino]

Dear Bas,

Il 28/11/25 10:29, Sebastiaan Couwenberg ha scritto:
> On 11/28/25 10:09 AM, Antonio Valentino wrote:
>> Il 27/11/25 08:32, Sebastiaan Couwenberg ha scritto:
>>> On 11/23/25 10:52 AM, Sebastiaan Couwenberg wrote:
>>>> The recent upload FTBFS on armhf which will need to be fixed before 
>>>> any upload to unstable.
>>>>
>>>> It doesn't seem to be caused by your patch, it might be caused by 
>>>> the cross building patch or by changes in the toolchain.
>>>>
>>>> Will need to look into this on the porterbox when I have time.
>>>
>>> None of the patches are the cause of the FTBFS on armhf, changes in 
>>> GCC 15 are.
>>>
>>> I've updated the package to use GCC 14 on armhf as a workaround for 
>>> the time being.
>>>
>>> There has been no response ARM porters to my request for help, that's 
>>> discouraging.
>>>
>>> We may need to exclude hdf4 as a dependency for gdal on armhf to 
>>> enable the removal of libhdf4 and its rdeps on armhf when gcc-14 is 
>>> removed from testing.
>>
>> I hope that there is still some time before gcc-14 removal.
> 
> Seeing that gcc-12 is still in stable & testing, I suspect that armhf 
> will be dropped as a release architecture before gcc-14 is removed from 
> testing.

Understood

> I would certainly be in favor of getting rid armhf.
> 
>> Maybe porters will find a way to solve the issue before the removal.
>>
>> Also in the HDFGroup side I see no reaction to our bug report [1].
> 
> Not unexpected, their focus is on HDF5.

I see.
They are in the process of releasing HDF5 v2.0 in these days

>> This means that we will have python-rioxarray and sarsen removed from 
>> testing tomorrow in addition to satpy, xsar and xarray-safe-rcm that 
>> have been already removed and are unable to migrate.
> 
> Fortunately testing removals are only temporary, and there is plenty of 
> time until the forky freeze.

Yes, I know ... but I would prefer to avid to have the packages removed 
form testing anyway.

>> Any option?
> 
> As someone who doesn't actually use hdf4 I don't have strong feelings 
> about this package.
> 
> Moving the updated package to unstable seems like the best way forward, 
> we don't have other fixes for the segfaults.

That would be nice.
Thanks for your help.

>> Do you have any direct contact with HDF4 upstream developer to try to 
>> understand if they have a plan for the near future about our issue?
> 
> The HDFGroup used Jira in the past, but now that they are using GitHub 
> that is much more convenient.
> 
> There are still links to the Help Desk on https://support.hdfgroup.org/ 
> hdf4/
> 
> You can try getting their attention by creating a Jira issue for HDF4.

OK, I will try


kind regards
-- 
Antonio Valentino