[Debichem-devel] Bug#982567: openms build-depends on removed package

Filippo Rusconi lopippo at debian.org
Tue Mar 2 08:16:46 GMT 2021


Greetings, Adrian,

On Mon, Mar 01, 2021 at 02:32:03AM +0200, Adrian Bunk wrote:
>On Sat, Feb 20, 2021 at 05:03:45PM +0100, Filippo Rusconi wrote:
>> Greetings, Andreas and Michael,
>>
>> I hope you are doing fine.
>>
>> On Fri, Feb 19, 2021 at 10:19:11PM +0100, Andreas Tille wrote:
>> > Hi Filippo,
>> >
>> > this is extremely unfortunate.  However, I guess the alternative would
>> > have been to keep some RC buggy seqan-dev which would not have helped
>> > openms as well.  I tried the same as Peter and replaced the
>> > Build-Depends seqan-dev by libseqan2-dev.
>> >
>> > I can confirm the observation from Peter about the missing header file.
>> > I simply tried to comment those missing headers (next one is also
>> > missing):
>> >
>> >
>> > // #include <seqan/seq_io/guess_stream_format.h>
>> > // #include <seqan/seq_io/read_fasta_fastq.h>
>> [...]
>>
>> Thank you for your trials and explanations. I am no on vacation, which means
>> I'll finally find some time to try understand the implications of that SeqAn
>> upgrade.
>>
>> I'll keep you posted with my findings.
>
>The brute force approach works for me:
>1. install seqan-dev from buster (for step 2)
>2. cp -a /usr/include/seqan debian/
>3. in debian/control remove the seqan-dev build dependency
>4. in debian/rules pass -DSEQAN_INCLUDE_DIR=$(DEBIAN_DIR)
>   to dh_auto_configure

I first tried to check if putting some header files in the local source tree
would do. But no, in fact the file handling-related headers pull down almost all
the seqan headers, so this strategy did not work out, leading me to envision
exactly what you tried.  I now wonder where to put the seqan headers when we
install the -dev stuff, so that people working with OpenMS do find them. I
thought creating a seqan subdirectory to the /usr/include/OpenMS directory and
point the compiler to that location by providing -I/usr/include/OpenMS.

But then, how do we inform the users about the unconventional location of the
header files? Do they actually need that location ?

Any idea ?

>cu Adrian

Cheers,
Filippo

-- 

⢀⣴⠾⠻⢶⣦⠀  Filippo Rusconi, PhD
⣾⠁⢠⠒⠀⣿⡁   Research scientist at CNRS
⢿⡄⠘⠷⠚⠋⠀   Debian Developer
⠈⠳⣄⠀⠀⠀⠀  http://msxpertsuite.org
           http://www.debian.org



More information about the Debichem-devel mailing list